Clinical impact of early-onset sepsis on outcomes of necrotizing enterocolitis: a retrospective case-control study Authors: Wei Feng, Jinping Hou, Xiaohong Die, Zhenhua Guo, Wei Liu, Xuan Zhai, Yi Wang Published: 2026 ABSTRACT PurposeEarly-onset sepsis (EOS) is a significant issue associated with high morbidity and mortality in newborns, yet it has not received adequate attention among neonatal necrotizing enterocolitis (NEC). This study aimed to investigate the clinical characteristics of NEC patients with EOS and to assess the association between EOS and NEC-related outcomes.Patients and methods: The retrospective case–control study included 196 NEC patients (Bell’s stage ≥ II) admitted to a tertiary neonatal intensive care unit from January 2023 to March 2025. Of these, 98 patients with EOS were matched to 98 controls (without EOS) based on birth weight, gestational age, and gender. The medical records of eligible patients were reviewed, and the clinical characteristics were systematically analyzed. Statistical analyses were performed using R software.ResultsThe cohort included 196 patients, with a median gestational age of 32.1 weeks (IQR: 29.5, 34.4) and a median birth weight of 1615.0 g (IQR: 1227.5, 2050.0). Compared with the controls, cases showed significantly higher inflammatory markers (CRP and procalcitonin), lower platelet count, and higher incidences of hypoproteinemia, shock, and coagulopathy (all P < 0.05). Furthermore, ventilator support, blood transfusion, using of inotropes (dopamine), and Bell stage III were significantly more frequent among cases. Both the rate of surgical intervention [67.35% vs. 22.45%] and mortality [29.59% vs. 8.16%] were markedly elevated in the cases than in the controls. In multivariate conditional Logistic regression analysis, EOS emerged as an independent risk factor for both NEC-related surgical intervention [odds ratio (95% confidence interval) = 3.853 (1.724–8.612)] and death [odds ratio (95% confidence interval) = 5.017 (2.083–12.087)].ConclusionThe clinical characteristics of medical conditions identified in patients with EOS were distinct from those observed in patients without EOS. Moreover, EOS was associated with increased odds of NEC-related surgical intervention and mortality. Introduction Neonatal necrotizing enterocolitis (NEC) is an acute and severe inflammatory disease of the gastrointestinal tract that primarily affects preterm infants with significant morbidity and mortality [1]. Despite advances in medical technology, NEC still poses a major challenge in neonatal care units [2]. The exact pathogenesis of NEC remains incompletely elucidated. Currently, it is widely acknowledged that the diease arises from the interplay of multiple factors, including immature gut barrier function, abnormal microbial colonization, improper feeding, and uncontrolled systemic inflammatory responses [3]. Within this intricate pathophysiological network, infection and systemic inflammation are recognized as pivotal drivers in advancing disease progression to intestinal necrosis [4–5]. Based on the timing of onset, early-onset sepsis (EOS) is usually defined as infection before 72 h of life, representing a significant threat to neonatal morbidity and mortality [6–7]. Given the early onset of sepsis, most studies suggest that pathogens are primarily transmitted vertically from mother to child during the pre- and perinatal periods [8]. EOS induces a severe systemic inflammatory response syndrome, resulting in multi-organ dysfunction [9]. Pathophysiologically, the systemic inflammation, hemodynamic disturbances, and coagulopathy caused by sepsis may exacerbate the pre-existing intestinal ischemia and inflammatory damage in NEC [4, 10, 11], thereby potentially influencing the severity and clinical outcomes of NEC. Although both NEC and EOS are independently associated with adverse outcomes in neonates, the interplay between the two conditions, particularly the impact of EOS on NEC-related outcomes, has not been fully elucidated. Consequently, this retrospective case-control study was designed to investigate the clinical characteristics of NEC complicated by EOS and to assess the association between EOS and NEC-related outcomes in a tertiary neonatal intensive care unit (NICU). By analyzing NEC patients with and without EOS, we hypothesized that NEC patients with EOS experienced more severe medical conditions and clinical outcomes. This findings may provide valuable insights for clinicians, enhancing the understanding of the intricate relationship between NEC and EOS, and ultimately contributing to improved outcomes for affected patients. Materials and methods Study design and setting This study was approved by the Institutional Research Ethics Board of the Children’s Hospital Affiliated Chongqing Medical University (Date: 2023/No: 01), and complied with the 1964 Helsinki Declaration. Because of the retrospective design, the requirement for parental informed consent was waived. In addition, all methods were performed in accordance with the Reporting of Cohort Studies in Surgery (STROCSS) criteria [12]. The medical records of patients with NEC admitted to Children’s Hospital of Chongqing Medical University from January 2023 to March 2025 were reviewed retrospectively. To mitigate potential confounding factors that could affect the validity of study, the following inclusion and exclusion criteria were applied. Inclusion criteria (1) patients diagnosed with NEC based on the criteria originally proposed by Bell et al. [13] (Bell’s stage ≥ II) [14], and (2) those with complete medical records and follow-up data. Exclusion criteria (1) patients with digestive system deformities (24 cases), like intestinal atresia, (2) instances where the patient’s guardian did not cooperate or opted to discontinue treatment (11 cases), (3) patients with genetic metabolic diseases (4 cases) or hematologic diseases (2 cases), and (4) those with incomplete clinical data (17 cases). Furthermore, patients with late-onset sepsis (191 cases) were intentionally excluded because EOS and late-onset sepsis are clinically distinct conditions with different temporal trajectories, likely pathogen profiles, and management pathways [6, 15], and this study was designed to assess the specific association between EOS exposure and NEC-related outcomes. Including patients with late-onset sepsis could attenuate the EOS-specific signal and introduce heterogeneity that would complicate causal interpretation. The entire number of patients who met the inclusion and exclusion criteria during the study time frame was 411 (Fig. 1), 98 of them were diagnosed with EOS (23.8%). For the accuracy of statistical analysis, controls were selected from this population and were matched for birth weight (± 50 g), gestational age (± 1 week), year of birth, and then gender at a 1:1 case: control ratio (individual matching). Fig. 1Flow chart of the study population Flow chart of the study population Data collection The collected data from medical records were categorized into five main areas, as follows.General demographic data: gender, gestational age, birth weight, Apgar score at 5 min, in vitro fertilization (IVF), small for gestational age (SGA) [16], fetal distress, premature rupture of membranes (PROM), meconium-staining amniotic fluid (MSAF), age at NEC onset, and congenital heart disease (CHD) [17];Maternal-related factors: maternal age, pregnancy, vaginal delivery, gestational hypertension (GH), gestational diabetes mellitus (GDM), intrahepatic cholestasis of pregnancy (ICP), and antenatal steroid exposure; Comorbidities and special treatments before NEC onset: hypoproteinemia [18], hyponatremia [19], coagulopathy [14], shock [17], ventilator support, blood transfusion, and using of inotropes (dopamine);General blood examination results measured at NEC onset: white blood cell count (WBC), neutrophil count (NE), hemoglobin (HB), platelet count (PLT), C-reactive protein (CRP), and procalcitonin.Outcomes: NEC-related treatment strategy (surgical or conservative) and prognosis (dead or survival) during hospitalization or within the 6-month follow-up period after discharge. General demographic data: gender, gestational age, birth weight, Apgar score at 5 min, in vitro fertilization (IVF), small for gestational age (SGA) [16], fetal distress, premature rupture of membranes (PROM), meconium-staining amniotic fluid (MSAF), age at NEC onset, and congenital heart disease (CHD) [17]; General demographic data: gender, gestational age, birth weight, Apgar score at 5 min, in vitro fertilization (IVF), small for gestational age (SGA) [16], fetal distress, premature rupture of membranes (PROM), meconium-staining amniotic fluid (MSAF), age at NEC onset, and congenital heart disease (CHD) [17]; Maternal-related factors: maternal age, pregnancy, vaginal delivery, gestational hypertension (GH), gestational diabetes mellitus (GDM), intrahepatic cholestasis of pregnancy (ICP), and antenatal steroid exposure; Maternal-related factors: maternal age, pregnancy, vaginal delivery, gestational hypertension (GH), gestational diabetes mellitus (GDM), intrahepatic cholestasis of pregnancy (ICP), and antenatal steroid exposure; Comorbidities and special treatments before NEC onset: hypoproteinemia [18], hyponatremia [19], coagulopathy [14], shock [17], ventilator support, blood transfusion, and using of inotropes (dopamine); Comorbidities and special treatments before NEC onset: hypoproteinemia [18], hyponatremia [19], coagulopathy [14], shock [17], ventilator support, blood transfusion, and using of inotropes (dopamine); General blood examination results measured at NEC onset: white blood cell count (WBC), neutrophil count (NE), hemoglobin (HB), platelet count (PLT), C-reactive protein (CRP), and procalcitonin. General blood examination results measured at NEC onset: white blood cell count (WBC), neutrophil count (NE), hemoglobin (HB), platelet count (PLT), C-reactive protein (CRP), and procalcitonin. Outcomes: NEC-related treatment strategy (surgical or conservative) and prognosis (dead or survival) during hospitalization or within the 6-month follow-up period after discharge. Outcomes: NEC-related treatment strategy (surgical or conservative) and prognosis (dead or survival) during hospitalization or within the 6-month follow-up period after discharge. Definitions For the diagnosis of sepsis, we referenced the expert consensus on the diagnosis and treatment of neonatal sepsis (China, 2019) [20]. EOS was defined as sepsis diagnosed within the first 72 h of life and antimicrobial treatment for at least 5 days [7]. Statistical analysis Excel software was used to data entry, and R software (version 4.3.2; https://www.r-project.org/) was used for statistical analysis. Categorical data were expressed by numbers (%), and compared using the Chi-square (χ2) or Fisher’s exact test. Continuous variables were tested for normality using Shapiro-Wilk test. Continuous variables without normal distribution were expressed as median and interquartile range (IQR) and compared using Mann–Whitney U test. Given the design features of small sample size, multiple variables, and multicollinearity among variables in this case–control study, independent factors were screened using the least absolute shrinkage and selection operator (LASSO) and multivariate conditional Logistic statistical method [21, 22]. First, LASSO (“glmnet” package) regression was utilized to determine the effects of potential factors on NEC-related surgical intervention and death, selecting the optimal tuning parameter at lambda.min. Then, these screened factors were entered into the multivariate conditional Logistic regression to identify independent predictors, reported as odds ratio (OR) with 95% confidence interval (CI). Variance inflation factors (VIFs) were examined to ensure the absence of collinearity between included variables. In statistical analysis, the test level was α = 0.05 (bilateral). Results Clinical characteristics of included patients 98 NEC patients with EOS were identified during the study time frame and matched with 98 controls (without EOS), providing a cohort of n = 196. All participants met the inclusion and exclusion criteria. The flow diagram for participants selection is shown in Fig. 1. The median gestational age was 32.1 weeks (IQR: 29.5, 34.4), and the median birth weight was 1615.0 g (IQR: 1227.5, 2050.0). The median interval between EOS diagnosis and NEC onset was 6.70 days (IQR: 3.85, 10.30), and all EOS patients were definitively diagnosed prior to NEC. Table 1 shows the demographic information of those participants. The median gestational age was 31.8 (IQR: 29.2, 34.4) weeks in the cases and 32.4 (IQR: 30.1, 34.4) weeks in the controls. The birth weights of patients in the cases (median: 1612.5 g) and controls (median: 1615.0 g) were comparable (P > 0.05). Furthermore, males constituted 43.88% of the cases, compared with 40.82% of the controls. In addition to matched characteristics (gestational age, birth weight, and gender), no statistically significant differences were observed between the two groups regarding IVF, SGA, Apgar score at 5 min, fetal distress, PROM, MSAF, CHD, pregnancy information, GH, GDM, ICP, pregnancy, vaginal delivery, or antenatal steroid exposure (all P > 0.05). However, cases had a significantly higher proportion of maternal age ≥ 35 years and experienced earlier age at NEC onset than controls (both P < 0.05). Table 1Analysis of baseline characteristics for NEC patients with and without EOSVariablesOverall patients(N = 196)NEC withoutEOS (N = 98)NEC withEOS (N = 98)P valueGeneral demographic dataGender (n/%)ǂ0.772 Male83 (42.35)40 (40.82)43 (43.88) Female113 (57.65)58 (59.18)55 (56.12)Gestational age (weeks, n/%)ǂ32.1 (29.5, 34.4)32.4 (30.1, 34.4)31.8 (29.2, 34.4)0.488Birth weight (grams, n/%)ǂ1615.0 (1227.5, 2050.0)1615.0 (1238.8, 2050.00)1612.5 (1230.0, 2036.3)0.756Test-tube baby (n/%)ǂ0.695 No165 (84.18)81 (82.65)84 (85.71) Yes31 (15.82)17 (17.35)14 (14.29)SGA (n/%)ǂ0.762 No65 (33.16)31 (31.63)34 (34.69) Yes131 (66.84)67 (68.37)64 (65.31)Apgar at 5 min (n/%)ǂ1.0 ≥ 7125 (63.78)62 (63.265)63 (64.29) <771 (36.22)36 (36.735)35 (35.71)Age at NEC onset (days, n/%)ǂ< 0.001 ≤ 759 (30.102)14 (14.29)45 (45.92) > 7 ~ 1473 (37.245)30 (30.61)43 (43.88) > 1464 (32.653)54 (55.10)10 (10.20)Fetal distress (n/%)ǂ0.657 No173 (88.27)85 (86.73)88 (89.80) Yes23 (11.73)13 (13.27)10 (10.20)PROM (n/%)ǂ0.180 No126 (64.29)68 (69.39)58 (59.18) Yes70 (35.71)30 (30.61)40 (40.82)MSAF (n/%)ǂ0.782 No182 (92.86)90 (91.84)92 (93.88) Yes14 (7.14)8 (8.16)6 (6.12)CDH (n/%)ǂ0.198 No96 (48.98)53 (54.08)43 (43.88) Yes100 (51.02)45 (45.92)55 (56.12)Pregnancy informationMaternal age (year, n/%)ǂ< 0.001 < 35120 (61.22)78 (79.59)42 (42.86) ≥ 3576 (38.78)20 (20.41)56 (57.14)GH (n/%)ǂ1.0 No172 (87.76)86 (87.76)86 (87.76) Yes24 (12.24)12 (12.24)12 (12.24)GDM (n/%)ǂ0.631 No142 (72.45)69 (70.41)73 (74.49) Yes54 (27.55)29 (29.59)25 (25.51)ICP (n/%)ǂ0.434 No180 (91.84)88 (89.80)92 (93.88) Yes16 (8.16)10 (10.20)6 (6.12)Pregnancy (n/%)ǂ0.060 Single114 (58.16)50 (51.02)64 (65.31) Multiple82 (41.84)48 (48.98)34 (34.69)Vaginal delivery (n/%)0.859 No40 (20.41)21 (21.43)19 (19.39) Yes156 (79.59)77 (78.57)79 (80.61)Antenatal steroid exposure (n/%)0.715 No37 (18.88)20 (20.41)17 (17.35) Yes159 (81.12)78 (79.59)81 (82.65)SGA small for gestational age, PROM premature rupture of membrane, MASF meconium-staining amniotic fluid, CHD congenital heart disease, GH gestational hypertension, GDM gestational diabetes mellitus, ICP intrahepatic cholestasis of pregnancy, EOS early-onset sepsisǂ Data are presented as n (%) and used Chi-square test#Values are presented as medians (IQR) and used Mann-Whitney U test Table 1Analysis of baseline characteristics for NEC patients with and without EOSVariablesOverall patients(N = 196)NEC withoutEOS (N = 98)NEC withEOS (N = 98)P valueGeneral demographic dataGender (n/%)ǂ0.772 Male83 (42.35)40 (40.82)43 (43.88) Female113 (57.65)58 (59.18)55 (56.12)Gestational age (weeks, n/%)ǂ32.1 (29.5, 34.4)32.4 (30.1, 34.4)31.8 (29.2, 34.4)0.488Birth weight (grams, n/%)ǂ1615.0 (1227.5, 2050.0)1615.0 (1238.8, 2050.00)1612.5 (1230.0, 2036.3)0.756Test-tube baby (n/%)ǂ0.695 No165 (84.18)81 (82.65)84 (85.71) Yes31 (15.82)17 (17.35)14 (14.29)SGA (n/%)ǂ0.762 No65 (33.16)31 (31.63)34 (34.69) Yes131 (66.84)67 (68.37)64 (65.31)Apgar at 5 min (n/%)ǂ1.0 ≥ 7125 (63.78)62 (63.265)63 (64.29) <771 (36.22)36 (36.735)35 (35.71)Age at NEC onset (days, n/%)ǂ< 0.001 ≤ 759 (30.102)14 (14.29)45 (45.92) > 7 ~ 1473 (37.245)30 (30.61)43 (43.88) > 1464 (32.653)54 (55.10)10 (10.20)Fetal distress (n/%)ǂ0.657 No173 (88.27)85 (86.73)88 (89.80) Yes23 (11.73)13 (13.27)10 (10.20)PROM (n/%)ǂ0.180 No126 (64.29)68 (69.39)58 (59.18) Yes70 (35.71)30 (30.61)40 (40.82)MSAF (n/%)ǂ0.782 No182 (92.86)90 (91.84)92 (93.88) Yes14 (7.14)8 (8.16)6 (6.12)CDH (n/%)ǂ0.198 No96 (48.98)53 (54.08)43 (43.88) Yes100 (51.02)45 (45.92)55 (56.12)Pregnancy informationMaternal age (year, n/%)ǂ< 0.001 < 35120 (61.22)78 (79.59)42 (42.86) ≥ 3576 (38.78)20 (20.41)56 (57.14)GH (n/%)ǂ1.0 No172 (87.76)86 (87.76)86 (87.76) Yes24 (12.24)12 (12.24)12 (12.24)GDM (n/%)ǂ0.631 No142 (72.45)69 (70.41)73 (74.49) Yes54 (27.55)29 (29.59)25 (25.51)ICP (n/%)ǂ0.434 No180 (91.84)88 (89.80)92 (93.88) Yes16 (8.16)10 (10.20)6 (6.12)Pregnancy (n/%)ǂ0.060 Single114 (58.16)50 (51.02)64 (65.31) Multiple82 (41.84)48 (48.98)34 (34.69)Vaginal delivery (n/%)0.859 No40 (20.41)21 (21.43)19 (19.39) Yes156 (79.59)77 (78.57)79 (80.61)Antenatal steroid exposure (n/%)0.715 No37 (18.88)20 (20.41)17 (17.35) Yes159 (81.12)78 (79.59)81 (82.65)SGA small for gestational age, PROM premature rupture of membrane, MASF meconium-staining amniotic fluid, CHD congenital heart disease, GH gestational hypertension, GDM gestational diabetes mellitus, ICP intrahepatic cholestasis of pregnancy, EOS early-onset sepsisǂ Data are presented as n (%) and used Chi-square test#Values are presented as medians (IQR) and used Mann-Whitney U test Analysis of baseline characteristics for NEC patients with and without EOS SGA small for gestational age, PROM premature rupture of membrane, MASF meconium-staining amniotic fluid, CHD congenital heart disease, GH gestational hypertension, GDM gestational diabetes mellitus, ICP intrahepatic cholestasis of pregnancy, EOS early-onset sepsis ǂ Data are presented as n (%) and used Chi-square test #Values are presented as medians (IQR) and used Mann-Whitney U test Differences in clinical features between patients with and without EOS are described in Table 2. The most common comorbidities were hypoproteinemia (42.35%, n = 83), followed by coagulopathy (33.67%, n = 66) and hyponatremia (32.65%, n = 64). Compared with controls, cases showed significantly higher inflammatory markers (CRP and procalcitonin), lower PLT, and higher incidences of hypoproteinemia, shock, and coagulopathy (all P < 0.05). Furthermore, ventilator support, blood transfusion, using of inotropes (dopamine), and Bell stage III were significantly more frequent among cases (all P < 0.05). No statistically significant differences were observed between the two groups in hyponatremia, WBC, HB, LY, or NE (all P > 0.05). Table 2Clinical features of NEC patients with and without EOSVariablesOverall patients(N = 196)NEC withoutEOS (N = 98)NEC withEOS (N = 98)P valueComorbidities and special treatments before NEC onsetHypoproteinemia (n/%)ǂ0.001 No113 (57.65)69 (70.41)44 (44.90) Yes83 (42.35)29 (29.59)54 (55.10)Hyponatremia (n/%)ǂ0.286 No132 (67.35)70 (71.43)62 (63.27) Yes64 (32.65)28 (28.57)36 (36.73)Coagulopathy (n/%)ǂ0.001 No130 (66.33)77 (78.57)53 (54.08) Yes66 (33.67)21 (21.43)45 (45.92)Shock (n/%)ǂ0.008 No156 (79.59)86 (87.76)70 (71.43) Yes40 (20.41)12 (12.24)28 (28.57)Ventilator support (n/%)ǂ0.017 No127 (64.80)72 (73.47)55 (56.12) Yes69 (35.20)26 (26.53)43 (43.88)Blood transfusion (n/%)ǂ0.005 No148 (75.51)83 (84.69)65 (66.33) Yes48 (24.49)15 (15.31)33 (33.67)Using of inotropes (dopamine) (n/%)ǂ0.011 No128 (65.31)73 (74.49)55 (56.12) Yes68 (34.69)25 (25.51)43 (43.88)General blood examination results measured at NEC onset WBC (×109/L)#8.9 (6.0, 13.4)8.4 (5.1, 12.6)9.5 (6.5, 13.8)0.078 HB (g/L)#129.5 (111.0, 152.0)127.5 (109.0, 151.0)131.5 (114.0, 153.0)0.391 LY (×109/L)2.8 (1.4, 4.2)2.7 (1.4, 4.1)2.9 (1.3, 4.4)0.580 NE (×109/L)#4.6 (2.9, 6.8)4.6 (2.9, 6.7)4.6 (2.9, 6.9)0.415 PLT (×109/L)#220.0 (114.5, 292.3)243.5 (146.8, 324.0)172.5 (91.3, 278.8)0.004CRP (mg/L, n/%)ǂ< 0.001 < 823 (11.73)20 (20.41)3 (3.06) 8 ~ 2089 (45.41)51 (52.04)38 (38.78) > 20 ~ 5041 (20.92)18 (18.37)23 (23.47) > 5043 (21.94)9 (9.18)34 (34.69)Procalcitonin (ng/ml)#2.9 (0.4, 11.2)2.0 (0.3, 7.9)5.8 (0.5, 13.6)0.018Bell’s stage (n/%)ǂ< 0.001 Ⅱ141 (71.94)82 (83.67)59 (60.20) Ⅲ55 (28.06)16 (16.33)39 (39.80)WBC white blood cell count, NE neutrophil count, LY lymphocyte count, HB hemoglobin, PLT platelet count, CRP C-reactive protein, EOS early-onset sepsisǂ Data are presented as n (%) and used Chi-square test#Values are presented as medians (IQR) and used Mann-Whitney U test Table 2Clinical features of NEC patients with and without EOSVariablesOverall patients(N = 196)NEC withoutEOS (N = 98)NEC withEOS (N = 98)P valueComorbidities and special treatments before NEC onsetHypoproteinemia (n/%)ǂ0.001 No113 (57.65)69 (70.41)44 (44.90) Yes83 (42.35)29 (29.59)54 (55.10)Hyponatremia (n/%)ǂ0.286 No132 (67.35)70 (71.43)62 (63.27) Yes64 (32.65)28 (28.57)36 (36.73)Coagulopathy (n/%)ǂ0.001 No130 (66.33)77 (78.57)53 (54.08) Yes66 (33.67)21 (21.43)45 (45.92)Shock (n/%)ǂ0.008 No156 (79.59)86 (87.76)70 (71.43) Yes40 (20.41)12 (12.24)28 (28.57)Ventilator support (n/%)ǂ0.017 No127 (64.80)72 (73.47)55 (56.12) Yes69 (35.20)26 (26.53)43 (43.88)Blood transfusion (n/%)ǂ0.005 No148 (75.51)83 (84.69)65 (66.33) Yes48 (24.49)15 (15.31)33 (33.67)Using of inotropes (dopamine) (n/%)ǂ0.011 No128 (65.31)73 (74.49)55 (56.12) Yes68 (34.69)25 (25.51)43 (43.88)General blood examination results measured at NEC onset WBC (×109/L)#8.9 (6.0, 13.4)8.4 (5.1, 12.6)9.5 (6.5, 13.8)0.078 HB (g/L)#129.5 (111.0, 152.0)127.5 (109.0, 151.0)131.5 (114.0, 153.0)0.391 LY (×109/L)2.8 (1.4, 4.2)2.7 (1.4, 4.1)2.9 (1.3, 4.4)0.580 NE (×109/L)#4.6 (2.9, 6.8)4.6 (2.9, 6.7)4.6 (2.9, 6.9)0.415 PLT (×109/L)#220.0 (114.5, 292.3)243.5 (146.8, 324.0)172.5 (91.3, 278.8)0.004CRP (mg/L, n/%)ǂ< 0.001 < 823 (11.73)20 (20.41)3 (3.06) 8 ~ 2089 (45.41)51 (52.04)38 (38.78) > 20 ~ 5041 (20.92)18 (18.37)23 (23.47) > 5043 (21.94)9 (9.18)34 (34.69)Procalcitonin (ng/ml)#2.9 (0.4, 11.2)2.0 (0.3, 7.9)5.8 (0.5, 13.6)0.018Bell’s stage (n/%)ǂ< 0.001 Ⅱ141 (71.94)82 (83.67)59 (60.20) Ⅲ55 (28.06)16 (16.33)39 (39.80)WBC white blood cell count, NE neutrophil count, LY lymphocyte count, HB hemoglobin, PLT platelet count, CRP C-reactive protein, EOS early-onset sepsisǂ Data are presented as n (%) and used Chi-square test#Values are presented as medians (IQR) and used Mann-Whitney U test Clinical features of NEC patients with and without EOS WBC white blood cell count, NE neutrophil count, LY lymphocyte count, HB hemoglobin, PLT platelet count, CRP C-reactive protein, EOS early-onset sepsis ǂ Data are presented as n (%) and used Chi-square test #Values are presented as medians (IQR) and used Mann-Whitney U test Comparison of NEC-related treatment strategy and prognosis Among the 196 patients, 88 (44.90%) underwent NEC-related surgical intervention, and 38 deaths (18.88%) were reported. Notably, both the rate of surgical intervention [67.35% (66/98) vs. 22.45% (22/98), Fig. 2A] and mortality [29.59% (29/98) vs. 8.16% (8/98), Fig. 2B] were significantly higher in cases than in controls (both P < 0.05). Fig. 2Comparison of NEC-related outcomes. A Patients with EOS had higher rate of surgical intervention [67.35% (66/98) vs. 22.45% (22/98), χ²= 39.926, P < 0.001] than in those without; B the mortality of patients with EOS [29.59% (29/98) vs. 8.16% (8/98), χ²= 14.693, P < 0.001] were higher than that in patients without EOS Comparison of NEC-related outcomes. A Patients with EOS had higher rate of surgical intervention [67.35% (66/98) vs. 22.45% (22/98), χ²= 39.926, P < 0.001] than in those without; B the mortality of patients with EOS [29.59% (29/98) vs. 8.16% (8/98), χ²= 14.693, P < 0.001] were higher than that in patients without EOS Risk factors associated with NEC-related surgery In the LASSO regression, a ten-fold cross-validation was conducted to identify the optimal penalty parameter (lambda. min) for this model and detected 11 variables with nonzero coefficients: EOS, gestational age, CHD, IVF, ICP, GH, hypoproteinemia, coagulopathy, using of inotropes (dopamine), PLT, and CRP (Fig. 3). Then, these factors were incorporated into the the multivariate conditional Logistic regression analysis, which revealed the following independent factors for NEC-related surgical intervention (Table 3): gestational age [OR (95% CI) = 0.858 (0.759–0.971)], GH [OR (95% CI) = 3.711 (1.139–12.090)], hypoproteinemia [OR (95% CI) = 2.700 (1.231–5.925)], coagulopathy [OR (95% CI) = 73.616 (1.601–8.170)], PLT [OR (95% CI) = 0.996 (0.992–0.999)], CRP [reference: < 8 mg/L; > 50 mg/L, OR (95% CI) = 5.737 (1.289–25.537)], and EOS [OR (95% CI) = 3.853 (1.724–8.612)]. The assessment of collinearity among the these variables was conducted, yielding VIFs of 1.137, 1.013, 1.147, 1.153, 1.068, 1.199, and 1.340, respectively, indicating that there was no multiple collinearity relationship existed. Fig. 3Screening of variables based on LASSO regression for NEC-related surgery. A LASSO coefficients produced by the regression analysis. Coefficient profiles of 32 features plotted against log(λ), and as the penalty increases, less informative variables shrink toward zero. B LASSO coefficient profiles of the 11 variables. Ten-fold cross-validation for tuning parameter selection: the left dashed line indicates the value of λ with minimum mean squared error; the right dashed line represents the largest λ within 1 standard error of the minimum. C The most predictive subset of the feature. LASSO, least absolute shrinkage and selection operator Screening of variables based on LASSO regression for NEC-related surgery. A LASSO coefficients produced by the regression analysis. Coefficient profiles of 32 features plotted against log(λ), and as the penalty increases, less informative variables shrink toward zero. B LASSO coefficient profiles of the 11 variables. Ten-fold cross-validation for tuning parameter selection: the left dashed line indicates the value of λ with minimum mean squared error; the right dashed line represents the largest λ within 1 standard error of the minimum. C The most predictive subset of the feature. LASSO, least absolute shrinkage and selection operator Table 3Multivariate conditional Logistic regression analysis for independent risk factor of surgeryVariablesMultivariateβSEWaldOR (95%CI) P Gestational age (weeks)-0.1530.0635.9100.858 (0.759–0.971) 0.015 GH (Yes)1.3110.6034.7363.711 (1.139–12.090) 0.030 Hypoproteinemia (Yes)0.9930.4016.1382.700 (1.231–5.925) 0.013 Coagulopathy (Yes)1.2850.4169.5573.616 (1.601–8.170) 0.002 PLT (×109/L)-0.0040.0027.2110.996 (0.992–0.999) 0.007 CRP (mg/L)*12.966 0.005 8 ~ 20-0.1490.6520.0520.861 (0.240–3.093)0.819 > 20 ~ 500.5520.7100.6031.736 (0.432–6.984)0.437 > 501.7470.7625.2575.737 (1.289–25.537) 0.022 EOS (Yes)1.3490.41010.8023.853 (1.724–8.612) 0.001 GH gestational hypertension, PLT platelet count, CRP C-reactive protein, EOS Early-onset sepsis, β regression coefficient, SE standard error, OR odds ratio, 95% CI 95% confidence interval*Setting of dummy variables in ordered rank data: CRP at onset, < 8 mg/LThe P values with boldface indicate statistical significance Table 3Multivariate conditional Logistic regression analysis for independent risk factor of surgeryVariablesMultivariateβSEWaldOR (95%CI) P Gestational age (weeks)-0.1530.0635.9100.858 (0.759–0.971) 0.015 GH (Yes)1.3110.6034.7363.711 (1.139–12.090) 0.030 Hypoproteinemia (Yes)0.9930.4016.1382.700 (1.231–5.925) 0.013 Coagulopathy (Yes)1.2850.4169.5573.616 (1.601–8.170) 0.002 PLT (×109/L)-0.0040.0027.2110.996 (0.992–0.999) 0.007 CRP (mg/L)*12.966 0.005 8 ~ 20-0.1490.6520.0520.861 (0.240–3.093)0.819 > 20 ~ 500.5520.7100.6031.736 (0.432–6.984)0.437 > 501.7470.7625.2575.737 (1.289–25.537) 0.022 EOS (Yes)1.3490.41010.8023.853 (1.724–8.612) 0.001 GH gestational hypertension, PLT platelet count, CRP C-reactive protein, EOS Early-onset sepsis, β regression coefficient, SE standard error, OR odds ratio, 95% CI 95% confidence interval*Setting of dummy variables in ordered rank data: CRP at onset, < 8 mg/LThe P values with boldface indicate statistical significance Multivariate conditional Logistic regression analysis for independent risk factor of surgery GH gestational hypertension, PLT platelet count, CRP C-reactive protein, EOS Early-onset sepsis, β regression coefficient, SE standard error, OR odds ratio, 95% CI 95% confidence interval *Setting of dummy variables in ordered rank data: CRP at onset, < 8 mg/L The P values with boldface indicate statistical significance Risk factors associated with NEC-related death Similarly, LASSO regression showed that gestational age, Apgar score at 5 min, and EOS were potentially associated with NEC-related death (Fig. 4). The LASSO regression identified potential variables, which were then analyzed using multivariate conditional Logistic regression to adjust for covariates. And the results found that gestational age [OR (95% CI) = 0.789 (0.695–0.896)] and EOS [OR (95% CI) = 5.017 (2.083–12.087)] were independent factors of NEC-related death (Table 4). Fig. 4Screening of variables based on LASSO regression for NEC-related death. A LASSO coefficients produced by the regression analysis. Coefficient profiles of 32 features plotted against log(λ), and as the penalty increases, less informative variables shrink toward zero. B LASSO coefficient profiles of the 3 variables. Ten-fold cross-validation for tuning parameter selection: the left dashed line indicates the value of λ with minimum mean squared error; the right dashed line represents the largest λ within 1 standard error of the minimum. C The most predictive subset of the feature. LASSO, least absolute shrinkage and selection operator Screening of variables based on LASSO regression for NEC-related death. A LASSO coefficients produced by the regression analysis. Coefficient profiles of 32 features plotted against log(λ), and as the penalty increases, less informative variables shrink toward zero. B LASSO coefficient profiles of the 3 variables. Ten-fold cross-validation for tuning parameter selection: the left dashed line indicates the value of λ with minimum mean squared error; the right dashed line represents the largest λ within 1 standard error of the minimum. C The most predictive subset of the feature. LASSO, least absolute shrinkage and selection operator Table 4Multivariate conditional logistic regression analysis for independent risk factor of deathVariablesMultivariateβSEWaldOR (95% CI) P Gestational age (weeks)-0.2370.06513.4380.789 (0.695–0.896) < 0.001 EOS (Yes)1.6130.44912.9275.017 (2.083–12.087) < 0.001 EOS early-onset sepsis, C-reactive protein β regression coefficient, SE standard error, OR odds ratio, 95% CI 95% confidence intervalThe P values with boldface indicate statistical significance Table 4Multivariate conditional logistic regression analysis for independent risk factor of deathVariablesMultivariateβSEWaldOR (95% CI) P Gestational age (weeks)-0.2370.06513.4380.789 (0.695–0.896) < 0.001 EOS (Yes)1.6130.44912.9275.017 (2.083–12.087) < 0.001 EOS early-onset sepsis, C-reactive protein β regression coefficient, SE standard error, OR odds ratio, 95% CI 95% confidence intervalThe P values with boldface indicate statistical significance Multivariate conditional logistic regression analysis for independent risk factor of death EOS early-onset sepsis, C-reactive protein β regression coefficient, SE standard error, OR odds ratio, 95% CI 95% confidence interval The P values with boldface indicate statistical significance Discussion It’s known that both EOS and NEC are serious diseases with high morbidity and mortality in neonates, and a close interrelationship between the two has been documented [23, 24]. EOS may increase the risk of NEC by inducing intestinal inflammation and microbiota dysbiosis, which can subsequently lead to systemic immunosuppression and exacerbate the progression of sepsis [24, 25]. Furthermore, Huang et al. has identified shared immune-related biomarkers for both neonatal sepsis and NEC, including MAP2K6, CHKA, CA4, and DEFA3, indicating potential common inflammatory pathways [26]. The existing studies predominantly focus on the independent outcomes of NEC or EOS, with limited published reports on the clinical outcomes of NEC patients with EOS. Therefore, we designed this study to assess the specific association between EOS exposure and NEC-related outcomes, with the aim of enhancing early identification capabilities, optimizing treatment strategies, and ultimately improving the clinical outcomes of these patients. This retrospective case–control study provides compelling evidence that EOS is not only prevalent among NEC patients but also significantly exacerbates disease severity and adversely affects clinical outcomes. Concretely speaking, compared with patients without EOS, those with EOS exhibited more severe systemic inflammatory responses (CRP and procalcitonin), as well as more extensive metabolic and circulatory dysfunction (hypoproteinemia, shock, and coagulopathy). Furthermore, these patients required more intensive life support interventions, including ventilator support, blood transfusion and using of inotropes (dopamine). Crucially, after adjusting for confounding variables, EOS emerged as an independent risk factor for both surgical intervention (OR = 3.853, 95% CI = 1.724–8.612) and mortality (OR = 5.017, 95% CI = 2.083–12.087) in this cohort. These findings highlight the critical interplay between EOS and NEC, emphasizing the need for heightened clinical vigilance and targeted therapeutic strategies. The observed variations in inflammatory markers between the two groups align with the pathophysiological hypothesis that EOS primes and exacerbates systemic inflammatory responses. It’s well established that PLT, CRP, and procalcitonin are critical markers for evaluating the severity of inflammatory responses [27, 28]. In this study, elevated levels of CRP and procalcitonin, coupled with a decreased PLT, reflected a pronounced and potentially dysregulated immune activation, which may contribute to the exacerbation of intestinal injury through increased cytokine release and endothelial damage [18, 24, 29]. This systemic inflammatory milieu likely predisposes neonates to a severe form of NEC, characterized by rapid progression and multi-organ involvement. The increased incidence of hypoproteinemia, shock, and coagulopathy in the cases further corroborates the hypothesis that EOS imposes sepsis-induced tissue and organ damage onto the NEC pathophysiology, creating a synergistic effect that challenges systemic stability [9, 24]. In addition, we found that NEC patients with EOS were more likely administered ventilator support, blood transfusion, and using of inotropes (dopamine). Therefore, clinicians should anticipate increased resource utilization for NEC patients with EOS, thereby informing the strategic allocation of medical resources to high-risk patient populations. This study demonstrates that EOS was associated with increased odds of NEC-related surgical intervention. Specifically, EOS may not merely coexist with NEC but may actively promote disease progression to transmural necrosis or perforation. Systemic inflammatory response induced by EOS that can exacerbate intestinal mucosal injury, compromise perfusion, and facilitate bacterial translocation [9, 11, 30]. This biological synergy with NEC pathophysiology plausibly accelerates the severity of disease. Moreover, these pathways collectively create a “vicious cycle” where sepsis and intestinal necrosis fuel each other [24], resulting in the need for surgical intervention. Our findings emphasize the necessity for increased vigilance in NEC patients with EOS. Early recognition of this high-risk phenotype may aid in stratifying patients for more intensive monitoring, timely imaging, and proactive surgical consultation. Similarly, EOS has been identified as an independent risk factor for mortality among NEC patients. NEC-related death is frequently attributed to sepsis, multi-organ failure, and the complications of extensive intestinal necrosis [31–34]. EOS may lead to increased mortality through the following pathological processes. NEC and EOS exert a synergistic effect, resulting in the release of substantial quantities of inflammatory mediators [9, 35, 36]. It may exacerbate the existing systemic inflammatory response and predisposes patients to multiple organ failure, a well-established predictor of adverse outcomes in NEC [14, 17]. Our findings corroborate this assertion, as patients with EOS exhibited a higher incidences of shock, coagulopathy, and hypoproteinemia compared with those without EOS, indicating compromised organ function. Furthermore, the immunological exhaustion and metabolic derangements induced by EOS may impair the neonate’s ability to withstand the additional physiologic stress of NEC and its treatment (e.g. surgery). The elevated mortality risk reinforces the concept that the management of NEC must be integrated with aggressive EOS control. It also suggests that outcome benchmarks and risk prediction models for NEC should account for the presence of EOS as a critical prognostic variable. Strength and implication This study was conducted at a tertiary pediatric hospital (National Children’s Medical Center in China), utilizing a matched cohort from a single NICU to ensure homogeneity and standardability of disease in management protocols. The case-control design (matching on birth weight, gestational age, year of birth, and gender) effectively minimized the confounding effects of critical variables, thereby facilitating a clearer assessment of the specific impact of EOS. Furthermore, The use of LASSO and multivariate conditional Logistic statistical method further strengthened our ability to identify independent associations. These findings advocate for the integration of EOS evaluation into the standard assessment of NEC and suggest that NEC patients with EOS constitute a high-risk subgroup that necessitates intensified monitoring and potentially tailored therapeutic approaches. First, in any neonate diagnosed with EOS, a thorough and rigorous evaluation for NEC should be considered essential. Early identification of NEC could prompt a more aggressive medical management strategy aimed at controlling systemic infection and inflammation. Second, it is imperative to recognize NEC patients with EOS as a high-risk subgroup that necessitates enhanced surveillance for clinical deterioration, prompt escalation of supportive care, more frequent abdominal examinations and imaging, and early consultation with a pediatric surgical team. Furthermore, this study raises some important clinical questions, providing novel ideas for further research. First, it remains uncertain whether EOS increases the risk of NEC or if there are distinct differences in the pathogenesis between EOS-associated NEC and traditional NEC. Furthermore, the impact of EOS on the mid-to-long-term outcomes (e.g. brain injury, short bowel syndrome, and growth) of NEC patients remains inadequately understood. Limitation Nevertheless, this study is subject to several limitations. First, this analysis was retrospective and was conducted in a single hospital with a relatively small cohort, and the diversity of patients in terms of geographic, ethnic, and clinical practice characteristics was restricted. For example, as our hospital serves as the largest tertiary referral center for children in Southwest China, the cohort may comprise patients with relatively severe conditions. Second, the retrospective design inherently introduces biases, and the lack of standardized data collection may have led to the omission of other statistically significant factors in this analysis, e.g. feeding pattern and antibiotic exposure prior to NEC onset [37]. Third, patients with LOS were intentionally excluded because EOS and late-onset sepsis are clinically distinct conditions with different temporal trajectories. This approach may restrict external generalizability to the broader NEC population that includes both EOS and late-onset sepsis. Future studies should assess late-onset sepsis-related effects and potential interactions between sepsis timing and NEC outcomes. Conclusion To our knowledge, this is the first study to identify that EOS is a common and serious comorbidity in NEC patients, associated with a distinct and severe medical conditions. Furthermore, multivariate conditional Logistic regression analysis indicated that EOS was independently associated with increased odds of NEC-related surgical intervention and mortality. These findings may serve as a guide for clinical practice, thereby improving the prognosis of these patients. Future studies are needed to investigate the pathological mechanisms of EOS affecting clinical conditions and outcome of NEC patients. In addition, there is a need to develop prevention or intervention strategies to reduce the severity or risk of EOS to restore a balanced neonatal inflammatory response and prevent inflammation-associated intestinal damage.