Early prediction of severe RSV-associated ALRTI in Asian pediatric patients: a simple nomogram Authors: Lingling Tian, Xiaolong Hu, Yucui Qin, Liwen Li, Xueqin Xi, Tao Li, Yuna Bi Published: 2025 ABSTRACT ObjectAn effective predictive model composed of common clinical indices for assessing the risk of severe respiratory syncytial virus (RSV) associated acute lower respiratory tract infection (ALRTI) is crucial for timely intervention and improving prognosis. However, currently, this kind of model is limited.MethodTotal of 600 pediatric RSV participants, including 185 severe RSV-associated ALRTI cases, were enrolled. Univariate and multivariate logistic regressions identified independent predictors and developed a series of predictive models. The area under the receiver operating characteristic curve (AUROC), calibration plot, decision curve analysis, net reclassification index and integrated discrimination index were used to screen the best model, visualized as a nomogram. Stratification analysis by gender and RSV type was conducted to optimize the model’s clinical applicability, while both an internal and an external validation were used to verify its predictive efficiency.ResultAge, red blood cell count (RBC), alanine aminotransferase (ALT), and prematurity were identified as candidate predictors for severe RSV-associated ALRTI. Four predictive models were constructed in the training set, including AP (Age, Prematurity), APA (Age, Prematurity, ALT), ARA (Age, RBC, ALT), and ARAP (Age, RBC, ALT, Prematurity), with AUROC values of 0.827, 0.824, 0.826, and 0.828, respectively. ARAP achieved the highest AUROC in both internal (0.784) and external (0.770) validations. Using stratification analysis, the male RSV-A subgroup was identified as the optimum application scope of ARAP with the highest AUROC of 0.896.ConclusionARAP is a simple and effective tool for assessing the risk of severe RSV-associated ALRTI, especially in the male RSV-A subgroup.Clinical trial numberNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12887-025-06357-2. Introduction Severe respiratory syncytial virus (RSV) associated acute lower respiratory tract infection (ALRTI) can lead to wheezing, hypoxemia, need for intensive care and even death [1–7]. Early identification of severe RSV-associated ALRTI patients is crucial for timely intervention and improving prognosis of pediatric patients [8]. Currently, several factors, including young age, prematurity and comorbidity presence, were considered as predictors of severe RSV-associated ALRTI [9–12]. Based on these risk factors, researchers tried to develop predictive models incorporating clinical and laboratory indices to assist clinicians in identifying patients at heightened risk of severe RSV-associated ALRTI. Although external validation is still required, a predictive model including prematurity, weight on admission, respiratory rate, lymphocyte percentage, and outpatient use of glucocorticoids was developed and achieved an AUROC of 0.784 [13]. Another model comprising nine variables, including age, prematurity, comorbidities, seizures, neutrophil-to-lymphocyte ratio (NLR), interleukin-6 (IL-6), lactate dehydrogenase (LDH), D-dimer, and co-infection, was established with AUROC values of 0.865 and 0.822 in training and external validation cohorts [14]. However, predictive models including objective and relative few predictors are more practical for clinical use. A validated and more simplified predictive model is still needed. In this study, a simple nomogram was developed to assess and predict the risk of severe RSV-associated ALRTI in hospitalized pediatric patients in general ward, thereby supporting early diagnosis, facilitating timely intervention, and ultimately improving patient outcomes. Methods and materials The participants’ enrollment, grouping and the study design As shown in Fig. 1, a total of 662 pediatric participants with confirmed RSV infection were initially enrolled during hospitalization. 600 participants were included in the analysis ultimately. Fig. 1 Flow chart of participants’ enrollment, grouping, and the study design Flow chart of participants’ enrollment, grouping, and the study design The inclusion and exclusion criteria are as follow: the inclusion criteria were: (i) age between 28 days and 60 months; (ii) RSV infection was identified and subtyped into RSV-A and RSV-B using targeted next-generation sequencing (tNGS) of throat swabs [15]; (iii) complete medical records available during hospitalization. The exclusion criteria were: (i) presence of severe congenital heart disease or valvular heart disease; (ii) conditions associated with pulmonary hypertension (PAH), including idiopathic or hereditary PAH, interstitial lung disease, or thromboembolic PAH; (iii) bronchopulmonary dysplasia; (iv) incomplete medical records. According to the criteria established by the Respiratory Syncytial Virus Network (ReSViNET) in 2015 [16], RSV-associated ALRTI was defined as a clinical diagnosis of RSV bronchiolitis or RSV pneumonia. Severe RSV-associated ALRTI was defined as ALRTI presenting with signs of respiratory distress (e.g., tachypnea, retractions, nasal flaring, grunting), hypoxemia, or dehydration requiring intensive care support. Among the 600 enrolled participants, 464 were recruited from Shandong Provincial Maternal and Child Health Care Hospital between April 2023 and December 2024. Of these, 261 (56.25%) were enrolled during the winter months, coinciding with the typical RSV peak season. The 464 participants were randomly divided into a training set (n = 300) and an internal validation set (n = 164) in a 7:3 ratio [17, 18]. An additional 136 participants were enrolled from the First Affiliated Hospital of Nanchang University during the same period to serve as the external validation cohort. Ethics approval was taken from the Ethic committee of Shandong Provincial Maternal and Child Health Care Hospital with approval number 2025-011. Potential prediction variables A total of 39 variables, encompassing demographic information, physical examination findings, and laboratory indices, were collected from the electronic medical records of patients. The variables were detailed as follows: (1) Demographic information: age, gender, weight on admission, prematurity, breastfeeding status, presence of siblings, length of hospital stay, body temperature on admission, and duration of fever. (2) Physical examination findings: wheezing, tachypnea, inspiratory retraction, crackles, hypoxemia and convulsions. (3) Laboratory indices: RSV type, co-infection, differential blood cell count, NLR, platelet-to-lymphocyte ratio (PLR), monocyte-to-lymphocyte ratio (MLR), C-reactive protein (CRP), 25-hydroxyvitamin D [25(OH)D], albumin (Alb), alanine aminotransferase (ALT), LDH and creatine kinase-MB (CK-MB). Throat swabs were collected within 24 h of admission for RSV-A and RSV-B as well as other ninety-four common pathogens detection, including bacteria, viruses, and atypical pathogens, using tNGS. Co-infection was defined as RSV infection concurrent with one or more of the remaining pathogens. Statistical analysis Continuous variables were summarized as mean ± standard deviation (SD) or median with interquartile range (IQR), whereas categorical variables were described as percentages. Group comparisons were performed by the t-test, Mann-Whitney U test, one-way ANOVA, or Kruskal-Wallis H test for continuous variables, and the Chi-square or Fisher’s exact test for categorical variables. Collinearity among candidate variables in the training cohort was assessed primarily using the variance inflation factor (VIF). All demographic and laboratory variables were initially subjected to univariate logistic analysis to identify predictors in training set. Variables with statistically significant differences were subsequently included in multivariate logistic regression to identify independent predictors and construct risk predictive models. Model performance was evaluated using receiver operating characteristic curves (ROC) and validated across both internal and external validation cohorts. Calibration plots and decision curve analysis (DCA) were employed to assess predictive accuracy and clinical applicability. Furthermore, the net reclassification index (NRI) and integrated discrimination index (IDI) were computed to compare model discrimination. A nomogram was then developed based on the final logistic regression model using the Regression Modeling Strategies (rms) package in R. All statistical procedures were performed with R software (version 4.3.3; http://www.rproject.org/), and a two-tailed p < 0.05 was considered statistically significant. Results Characteristics of patients enrolled in this study Total of 300 participants were included in the training set, with a median age of 12 months. Among them, 90 were classified as severe RSV-associated ALRTI group and 210 were classified as the non-severe RSV-associated ALRTI group according to the ReSViNET in 2015 [16]. Participants with severe RSV-associated ALRTI had a lower median age [3.00 (IQR: 2.00, 8.75) vs. 19.00 (IQR: 8.00, 38.00), p < 0.001], lower body weight [6.82 (IQR: 5.50, 9.50) vs. 11.00 (IQR: 9.00, 15.00), p < 0.001], lower proportion of co-infection (27.8% vs. 51.9%, p < 0.001), and a higher proportion had a history of prematurity (12.2% vs. 4.8%, p = 0.038) or siblings (60.0% vs. 39.0%, p = 0.001). Additionally, these participants also showed significantly differences in the level of red blood cell (RBC) count and hemoglobin (Hb), platelet (PLT) count, mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), red cell distribution width (RDW), platelet distribution width (PDW), ALT, and CK-MB. These significant differences applied a foundation for identifying predictors of severe RSV-associated ALRTI. A similar distribution pattern was found in the external validation cohort, indicating balanced baseline characteristics between two cohorts. Detailed characteristics are summarized in Table 1 and Table S1. Table 1Characteristics of the pediatric patients with severe RSV-associated ALRTI and non-severe RSV-associated ALRTI groups in derivation cohortVariables (n = 39)Training set (n = 300)Internal validation set (n = 164)Non-severe RSV-associated ALRTI (n = 210)Severe RSV-associated ALRTI (n = 90) p Non-severe RSV-associated ALRTI (n = 114)Severe RSV-associated ALRTI (n = 50) p Age (IQR) (m)19.00 (8.00, 38.00)3.00 (2.00, 8.75)< 0.001*18.00 (8.00, 36.75)5.00 (3.25, 10.00)< 0.001*Gender, n (%)Female100 (47.6)35 (38.9)0.20544 (38.6)23 (46.0)0.474Male110 (52.4)55 (61.1)70 (61.4)27 (54.0)Weight (IQR) (kg)11.00 (9.00, 15.00)6.82 (5.50, 9.50)< 0.001*11.00 (8.62, 14.00)8.75 (6.50, 11.00)< 0.001*Prematurity, n (%)without200 (95.2)79 (87.8)0.038*107 (93.9)42 (84.0)0.085with10 (4.8)11 (12.2)7 (6.1)8 (16.0)Breastfeeding, n (%)without85 (40.50)42 (46.70)0.32041 (36)19 (38)0.803with125 (59.50)48 (53.30)73 (64)31 (62)Siblings, n (%)without128 (61.0)36 (40.0)0.001*70 (61.4)28 (56.0)0.634with82 (39.0)54 (60.0)44 (38.6)22 (44.0)LOS (IQR) (d)6.00 (5.00, 7.00)7.00 (6.00, 9.00)< 0.001*6.00 (5.00, 7.00)7.00 (6.00, 8.00)< 0.001*Temperature onadmission (IQR) (℃)36.50 (36.20, 36.90)36.60 (36.40, 36.80)0.03*36.70 (36.40, 37.08)36.50 (36.40, 36.80)0.184Fever duration (IQR) (d)2.00 (0.00, 4.00)0.00 (0.00, 1.00)< 0.001*2.00 (0.00, 3.75)1.00 (0.00, 4.00)0.514Wheezing, n (%)without133 (63.3)23 (25.6)< 0.001*64 (56.1)15 (30.0)0.004*with77 (36.7)67 (74.4)50 (43.9)35 (70.0)Tachypnea, n (%)without200 (95.2)25 (27.8)< 0.001*110 (96.5)25 (50.0)< 0.001*with10 (4.8)65 (72.2)4 (3.5)25 (50.0)Inspiratory retraction, n (%)without210 (100)23 (25.6)< 0.001*114 (100)24 (48.0)< 0.001*with067 (74.4)026 (52.0)Crackles, n (%)without76 (36.2)6 (6.7)< 0.001*37 (32.5)5 (10.0)0.005*with134 (63.8)84 (93.3)77 (67.5)45 (90.0)Hypoxemia, n (%)without210 (99.5)17 (18.9)< 0.001*114 (98.2)20 (40.0)< 0.001*with073 (81.1)030 (60.0)Convulsions, n (%)without209 (99.5)89 (98.9)1113 (99.1)49 (98.0)1with1 (0.5)1 (1.1)1 (0.9)1 (2.0)RSV type, n (%)Type A56 (26.70)20 (22.20)0.41714 (12.30)8 (16.00)0.520Type B154 (73.30)70 (77.80)100 (87.70)42 (84.00)Co-infectionwithout101 (48.1)65 (72.2)< 0.001*55 (48.2)36 (72.0)0.008*with109 (51.9)25 (27.8)59 (51.8)14 (28.0)WBC (IQR) (109/L)7.34 (5.93, 9.37)7.56 (5.73, 9.80)0.8547.16 (5.40, 8.72)7.94 (7.09, 9.22)0.02*RBC(SD) (1012/L)4.47 (0.40)3.98 (0.61)< 0.001*4.46 (0.40)4.09 (0.54)< 0.001*HB (IQR) (g/L)122.00 (114.00, 129.00)109.00 (100.25, 118.75)< 0.001*120.50 (13.30)115.04 (14.08)0.019*PLT (IQR) (109/L)304.50 (251.00, 360.50)370.00 (299.50, 437.50)< 0.001*279.00 (238.25, 358.50)355.50 (279.25, 425.00)0.002*MCV (SD) (fL)82.10 (4.07)84.54 (6.71)< 0.001*82.90 (79.70, 84.95)84.35 (81.25, 86.25)0.012*MCH (SD) (pg)27.21 (1.51)28.19 (2.53)< 0.001*27.40 (26.15, 28.20)27.85 (26.88, 29.10)0.018*MCHC (IQR) (g/L)332.00 (324.00, 338.00)331.00 (327.00, 340.00)0.198330.44 (9.06)325.06 (43.02)0.203NE (IQR) (109/L)2.34 (1.42, 3.87)1.91 (1.25, 2.99)0.1022.30 (1.49, 4.07)3.00 (1.84, 4.01)0.344EO (IQR) (109/L)0.06 (0.02, 0.15)0.06 (0.01, 0.17)0.9920.06 (0.01, 0.16)0.03 (0.01, 0.14)0.166LY (IQR) (109/L)3.96 (2.96, 5.30)4.24 (3.20, 5.62)0.3163.84 (2.48, 4.91)4.26 (3.32, 5.39)0.056MO (IQR) (109/L)0.57 (0.45, 0.76)0.65 (0.45, 0.97)0.1360.63 (0.42, 0.89)0.71 (0.55, 1.02)0.03*RDW (IQR)13.20 (12.80, 13.70)13.75 (13.00, 14.28)0.001*13.35 (12.70, 13.80)13.30 (12.62, 14.28)0.475PDW (IQR) (fL)15.65 (15.40, 15.90)15.80 (15.60, 16.20)< 0.001*15.70 (15.53, 15.97)15.90 (15.70, 16.17)0.027*NLR (IQR)0.57 (0.32, 1.08)0.44 (0.29, 0.80)0.0910.67 (0.41, 1.16)0.67 (0.37, 1.15)0.932PLR (IQR)79.44 (57.20, 101.16)89.56 (62.47, 112.55)0.05683.91 (58.00, 116.01)82.32 (61.40, 110.10)0.946MLR (IQR)0.14 (0.10, 0.21)0.15 (0.11, 0.20)0.5580.17 (0.12, 0.24)0.18 (0.13, 0.25)0.544CRP (IQR) (mg/L)1.46 (0.40, 5.07)0.73 (0.28, 2.27)0.0241.54 (0.37, 6.30)1.58 (0.39, 4.66)0.85825(OH)VitD (IQR) (ng/mL)37.61 (33.52, 40.65)37.61 (32.92, 38.92)0.9637.61 (35.78, 38.70)37.61 (36.42, 38.60)0.522Alb (SD) (g/L)40.46 (2.79)39.80 (3.63)0.08840.67 (3.98)40.02 (3.27)0.309ALT (IQR) (U/L)15.30 (13.00, 19.30)25.10 (18.40, 39.23)< 0.001*17.25 (13.33, 2.55)22.05 (17.58, 36.85)< 0.001*LDH (IQR) (U/L)298.65 (264.72, 336.60)309.97 (291.25, 309.97)0.094309.97 (294.65, 316.42)309.97 (261.92, 346.68)0.502CKMB (IQR) (ng/mL)1.96 (1.27, 3.14)4.06 (2.69, 5.63)< 0.001*2.24 (1.16, 3.81)3.12 (1.69, 4.99)0.019*LOS Length of hospital stays, RSV Respiratory syncytial virus, WBC White blood cell, RBC Red blood cell, HB Hemoglobin, PLT Platelet, MCV Mean corpuscular volume, MCH Mean corpuscular hemoglobin, MCHC Mean corpuscular hemoglobin concentration, NE Neutrophil, EO Eosinophil, LY Lymphocyte, MO Monocyte, RDW Red cell distribution width, PDW Platelet distribution width, NLR Neutrophil to lymphocyte ratio, PLR Platelet to lymphocyte ratio, MLR Monocyte to lymphocyte ratio, CRP c Reaction protein, 25(OH)VitD 25-hydroxy vitamin D, Alb Albumin, ALT Alanine aminotransferase, LDH Lactic dehydrogenase, CKMB Creatine kinase-MB, ALRTI Acute lower respiratory tract infectionData are expressed as mean ± SD or median (IQR) for continuous variables, and n (%) for categorical variables. Group comparisons were performed using the t-test or Mann–Whitney U test for continuous variables and the Chi-square test or Fisher’s exact test for categorical variables, as appropriate*p< 0.05 for significance Table 1Characteristics of the pediatric patients with severe RSV-associated ALRTI and non-severe RSV-associated ALRTI groups in derivation cohortVariables (n = 39)Training set (n = 300)Internal validation set (n = 164)Non-severe RSV-associated ALRTI (n = 210)Severe RSV-associated ALRTI (n = 90) p Non-severe RSV-associated ALRTI (n = 114)Severe RSV-associated ALRTI (n = 50) p Age (IQR) (m)19.00 (8.00, 38.00)3.00 (2.00, 8.75)< 0.001*18.00 (8.00, 36.75)5.00 (3.25, 10.00)< 0.001*Gender, n (%)Female100 (47.6)35 (38.9)0.20544 (38.6)23 (46.0)0.474Male110 (52.4)55 (61.1)70 (61.4)27 (54.0)Weight (IQR) (kg)11.00 (9.00, 15.00)6.82 (5.50, 9.50)< 0.001*11.00 (8.62, 14.00)8.75 (6.50, 11.00)< 0.001*Prematurity, n (%)without200 (95.2)79 (87.8)0.038*107 (93.9)42 (84.0)0.085with10 (4.8)11 (12.2)7 (6.1)8 (16.0)Breastfeeding, n (%)without85 (40.50)42 (46.70)0.32041 (36)19 (38)0.803with125 (59.50)48 (53.30)73 (64)31 (62)Siblings, n (%)without128 (61.0)36 (40.0)0.001*70 (61.4)28 (56.0)0.634with82 (39.0)54 (60.0)44 (38.6)22 (44.0)LOS (IQR) (d)6.00 (5.00, 7.00)7.00 (6.00, 9.00)< 0.001*6.00 (5.00, 7.00)7.00 (6.00, 8.00)< 0.001*Temperature onadmission (IQR) (℃)36.50 (36.20, 36.90)36.60 (36.40, 36.80)0.03*36.70 (36.40, 37.08)36.50 (36.40, 36.80)0.184Fever duration (IQR) (d)2.00 (0.00, 4.00)0.00 (0.00, 1.00)< 0.001*2.00 (0.00, 3.75)1.00 (0.00, 4.00)0.514Wheezing, n (%)without133 (63.3)23 (25.6)< 0.001*64 (56.1)15 (30.0)0.004*with77 (36.7)67 (74.4)50 (43.9)35 (70.0)Tachypnea, n (%)without200 (95.2)25 (27.8)< 0.001*110 (96.5)25 (50.0)< 0.001*with10 (4.8)65 (72.2)4 (3.5)25 (50.0)Inspiratory retraction, n (%)without210 (100)23 (25.6)< 0.001*114 (100)24 (48.0)< 0.001*with067 (74.4)026 (52.0)Crackles, n (%)without76 (36.2)6 (6.7)< 0.001*37 (32.5)5 (10.0)0.005*with134 (63.8)84 (93.3)77 (67.5)45 (90.0)Hypoxemia, n (%)without210 (99.5)17 (18.9)< 0.001*114 (98.2)20 (40.0)< 0.001*with073 (81.1)030 (60.0)Convulsions, n (%)without209 (99.5)89 (98.9)1113 (99.1)49 (98.0)1with1 (0.5)1 (1.1)1 (0.9)1 (2.0)RSV type, n (%)Type A56 (26.70)20 (22.20)0.41714 (12.30)8 (16.00)0.520Type B154 (73.30)70 (77.80)100 (87.70)42 (84.00)Co-infectionwithout101 (48.1)65 (72.2)< 0.001*55 (48.2)36 (72.0)0.008*with109 (51.9)25 (27.8)59 (51.8)14 (28.0)WBC (IQR) (109/L)7.34 (5.93, 9.37)7.56 (5.73, 9.80)0.8547.16 (5.40, 8.72)7.94 (7.09, 9.22)0.02*RBC(SD) (1012/L)4.47 (0.40)3.98 (0.61)< 0.001*4.46 (0.40)4.09 (0.54)< 0.001*HB (IQR) (g/L)122.00 (114.00, 129.00)109.00 (100.25, 118.75)< 0.001*120.50 (13.30)115.04 (14.08)0.019*PLT (IQR) (109/L)304.50 (251.00, 360.50)370.00 (299.50, 437.50)< 0.001*279.00 (238.25, 358.50)355.50 (279.25, 425.00)0.002*MCV (SD) (fL)82.10 (4.07)84.54 (6.71)< 0.001*82.90 (79.70, 84.95)84.35 (81.25, 86.25)0.012*MCH (SD) (pg)27.21 (1.51)28.19 (2.53)< 0.001*27.40 (26.15, 28.20)27.85 (26.88, 29.10)0.018*MCHC (IQR) (g/L)332.00 (324.00, 338.00)331.00 (327.00, 340.00)0.198330.44 (9.06)325.06 (43.02)0.203NE (IQR) (109/L)2.34 (1.42, 3.87)1.91 (1.25, 2.99)0.1022.30 (1.49, 4.07)3.00 (1.84, 4.01)0.344EO (IQR) (109/L)0.06 (0.02, 0.15)0.06 (0.01, 0.17)0.9920.06 (0.01, 0.16)0.03 (0.01, 0.14)0.166LY (IQR) (109/L)3.96 (2.96, 5.30)4.24 (3.20, 5.62)0.3163.84 (2.48, 4.91)4.26 (3.32, 5.39)0.056MO (IQR) (109/L)0.57 (0.45, 0.76)0.65 (0.45, 0.97)0.1360.63 (0.42, 0.89)0.71 (0.55, 1.02)0.03*RDW (IQR)13.20 (12.80, 13.70)13.75 (13.00, 14.28)0.001*13.35 (12.70, 13.80)13.30 (12.62, 14.28)0.475PDW (IQR) (fL)15.65 (15.40, 15.90)15.80 (15.60, 16.20)< 0.001*15.70 (15.53, 15.97)15.90 (15.70, 16.17)0.027*NLR (IQR)0.57 (0.32, 1.08)0.44 (0.29, 0.80)0.0910.67 (0.41, 1.16)0.67 (0.37, 1.15)0.932PLR (IQR)79.44 (57.20, 101.16)89.56 (62.47, 112.55)0.05683.91 (58.00, 116.01)82.32 (61.40, 110.10)0.946MLR (IQR)0.14 (0.10, 0.21)0.15 (0.11, 0.20)0.5580.17 (0.12, 0.24)0.18 (0.13, 0.25)0.544CRP (IQR) (mg/L)1.46 (0.40, 5.07)0.73 (0.28, 2.27)0.0241.54 (0.37, 6.30)1.58 (0.39, 4.66)0.85825(OH)VitD (IQR) (ng/mL)37.61 (33.52, 40.65)37.61 (32.92, 38.92)0.9637.61 (35.78, 38.70)37.61 (36.42, 38.60)0.522Alb (SD) (g/L)40.46 (2.79)39.80 (3.63)0.08840.67 (3.98)40.02 (3.27)0.309ALT (IQR) (U/L)15.30 (13.00, 19.30)25.10 (18.40, 39.23)< 0.001*17.25 (13.33, 2.55)22.05 (17.58, 36.85)< 0.001*LDH (IQR) (U/L)298.65 (264.72, 336.60)309.97 (291.25, 309.97)0.094309.97 (294.65, 316.42)309.97 (261.92, 346.68)0.502CKMB (IQR) (ng/mL)1.96 (1.27, 3.14)4.06 (2.69, 5.63)< 0.001*2.24 (1.16, 3.81)3.12 (1.69, 4.99)0.019*LOS Length of hospital stays, RSV Respiratory syncytial virus, WBC White blood cell, RBC Red blood cell, HB Hemoglobin, PLT Platelet, MCV Mean corpuscular volume, MCH Mean corpuscular hemoglobin, MCHC Mean corpuscular hemoglobin concentration, NE Neutrophil, EO Eosinophil, LY Lymphocyte, MO Monocyte, RDW Red cell distribution width, PDW Platelet distribution width, NLR Neutrophil to lymphocyte ratio, PLR Platelet to lymphocyte ratio, MLR Monocyte to lymphocyte ratio, CRP c Reaction protein, 25(OH)VitD 25-hydroxy vitamin D, Alb Albumin, ALT Alanine aminotransferase, LDH Lactic dehydrogenase, CKMB Creatine kinase-MB, ALRTI Acute lower respiratory tract infectionData are expressed as mean ± SD or median (IQR) for continuous variables, and n (%) for categorical variables. Group comparisons were performed using the t-test or Mann–Whitney U test for continuous variables and the Chi-square test or Fisher’s exact test for categorical variables, as appropriate*p< 0.05 for significance Characteristics of the pediatric patients with severe RSV-associated ALRTI and non-severe RSV-associated ALRTI groups in derivation cohort Temperature on admission (IQR) (℃) Fever duration (IQR) (d) LOS Length of hospital stays, RSV Respiratory syncytial virus, WBC White blood cell, RBC Red blood cell, HB Hemoglobin, PLT Platelet, MCV Mean corpuscular volume, MCH Mean corpuscular hemoglobin, MCHC Mean corpuscular hemoglobin concentration, NE Neutrophil, EO Eosinophil, LY Lymphocyte, MO Monocyte, RDW Red cell distribution width, PDW Platelet distribution width, NLR Neutrophil to lymphocyte ratio, PLR Platelet to lymphocyte ratio, MLR Monocyte to lymphocyte ratio, CRP c Reaction protein, 25(OH)VitD 25-hydroxy vitamin D, Alb Albumin, ALT Alanine aminotransferase, LDH Lactic dehydrogenase, CKMB Creatine kinase-MB, ALRTI Acute lower respiratory tract infection Data are expressed as mean ± SD or median (IQR) for continuous variables, and n (%) for categorical variables. Group comparisons were performed using the t-test or Mann–Whitney U test for continuous variables and the Chi-square test or Fisher’s exact test for categorical variables, as appropriate *p< 0.05 for significance Selection and evaluation of independent predictors in the training set Variables with severe multicollinearity (VIF ≥ 10), including MCV, MCH, MCHC, were excluded from subsequent regression analysis (Table S2). As shown in Table 2, 13 variables (age, prematurity, breastfeeding, presence of siblings, fever duration, co-infection, RBC, PLT, MO, RDW, PDW, Alb, and ALT) were identified as potential candidate predictors by univariate logistic regression in the training set. We further identified age (odds ratio [OR]: 0.963, 95% confidence interval [CI]: 0.922–0.998), RBC (OR: 0.243, 95% CI: 0.088–0.612), PDW (OR: 1.154, 95% CI: 0.976–1.381), and ALT (OR: 1.032, 95% CI: 1.004–1.067) as independent risk predictors for severe RSV-associated ALRTI by multivariate logistic regression. Table 2Univariate and multivariate analysis of risk factors for severe RSV-associated ALRTI in training setVariables(n = 13)UnivariateMultivariateOR95%CI p OR95%CI p Age (m)0.9150.887–0.9390.0001*0.9630.922–0.9980.064#Prematurity (%)2.7851.131–6.9360.025*Breastfeeding (%)0.5200.294–0.9230.025*Siblings (%)2.3411.419–3.9020.001*Fever duration (d)0.6080.496–0.7260.0001*Coinfection (%)0.3560.206–0.6020.0002*RBC (1012/L)0.1250.066–0.2250.0001*0.2430.088–0.6120.004*PLT (109/L)1.0061.003–1.0080.0001*MO (109/L)2.2611.044–4.9370.039*RDW (%)1.4401.121–1.8830.006*PDW (fL)1.2261.042–1.5660.048*1.1540.976–1.3810.065Alb (g/L)0.9300.854–1.010.089#ALT (U/L)1.0671.043–1.0940.0001*1.0321.004–1.0670.044*RBC Red blood cell, PLT Platelet, MO Monocyte, RDW Red cell distribution width, PDW Platelet distribution width, Alb Albumin, ALT Alanine aminotransferase, OR Odds ratio, CI Confidence interval#p< 0.1 for marginal significance Table 2Univariate and multivariate analysis of risk factors for severe RSV-associated ALRTI in training setVariables(n = 13)UnivariateMultivariateOR95%CI p OR95%CI p Age (m)0.9150.887–0.9390.0001*0.9630.922–0.9980.064#Prematurity (%)2.7851.131–6.9360.025*Breastfeeding (%)0.5200.294–0.9230.025*Siblings (%)2.3411.419–3.9020.001*Fever duration (d)0.6080.496–0.7260.0001*Coinfection (%)0.3560.206–0.6020.0002*RBC (1012/L)0.1250.066–0.2250.0001*0.2430.088–0.6120.004*PLT (109/L)1.0061.003–1.0080.0001*MO (109/L)2.2611.044–4.9370.039*RDW (%)1.4401.121–1.8830.006*PDW (fL)1.2261.042–1.5660.048*1.1540.976–1.3810.065Alb (g/L)0.9300.854–1.010.089#ALT (U/L)1.0671.043–1.0940.0001*1.0321.004–1.0670.044*RBC Red blood cell, PLT Platelet, MO Monocyte, RDW Red cell distribution width, PDW Platelet distribution width, Alb Albumin, ALT Alanine aminotransferase, OR Odds ratio, CI Confidence interval#p< 0.1 for marginal significance Univariate and multivariate analysis of risk factors for severe RSV-associated ALRTI in training set RBC Red blood cell, PLT Platelet, MO Monocyte, RDW Red cell distribution width, PDW Platelet distribution width, Alb Albumin, ALT Alanine aminotransferase, OR Odds ratio, CI Confidence interval #p< 0.1 for marginal significance The predictive performance of these identified independent predictors for severe RSV-associated ALRTI was assessed by AUROC. As shown in Fig. 2, the AUROC values for these independent predictors were as follows: Age 0.827 (95% CI: 0.773–0.882), RBC 0.737 (95% CI: 0.670–0.804), ALT 0.779 (95% CI: 0.722–0.836) and PDW 0.647 (95% CI: 0.581–0.714). Amongst, Age exhibited the highest predictive performance, followed by ALT, RBC, and PDW. Whereas PDW was not recommended for candidate predictors of subsequent model construction due to its low AUROC value obtained from training set. Fig. 2 Receiver operating characteristic curve (ROC) of four independent predictors for assessing the risk of severe RSV-associated ALRTI in training set. ROC of the Age (A), RBC (B), ALT (C), Prematurity (D) were plotted to verify the diagnostic accuracy of these predictors for assessing the risk of severe RSV-associated ALRTI in training set Receiver operating characteristic curve (ROC) of four independent predictors for assessing the risk of severe RSV-associated ALRTI in training set. ROC of the Age (A), RBC (B), ALT (C), Prematurity (D) were plotted to verify the diagnostic accuracy of these predictors for assessing the risk of severe RSV-associated ALRTI in training set Development of models for predicting severe RSV-associated ALRTI Prematurity, which has been widely reported as a significant predictor in many studies [9, 10, 12], was included as a candidate predictor in the construction of the predictive models. Four fitted and simplified predictive models, named as AP (including Age, Prematurity), APA (including Age, Prematurity, ALT), ARA (including Age, RBC, ALT), and ARAP (including Age, RBC, ALT, Prematurity) were constructed on the basis of these four candidate predictors. In the training set, all four models demonstrated good discriminatory performance in distinguishing severe RSV-associated ALRTI cases, with AUROC values as follows: AP 0.827 (95% CI: 0.773–0.882), APA 0.824 (95% CI: 0.770–0.879), ARA 0.826 (95% CI: 0.771–0.881), and ARAP 0.828 (95% CI: 0.772–0.883) (Fig. 3A-D). In the internal validation set, the AUROC values were AP 0.770 (95% CI: 0.692–0.855), APA 0.778 (95% CI: 0.704–0.853), ARA 0.778 (95% CI: 0.702–0.854), and ARAP 0.784 (95% CI: 0.709–0.860) (Fig. 3E-H). Sensitivity, specificity, positive predictive value, and negative predictive value of four fitted and simplified predictive models were summarized in Table S3. Fig. 3 ROC of 4 combined models for predicting the risk of severe RSV-associated ALRTI in training and internal validation sets. ROC of AP (A), APA (B), ARA (C) and ARAP (D) were plotted to verify the diagnostic accuracy of these models in training set. ROC of AP (E), APA (F), ARA (G) and ARAP (H) were plotted in internal validation set ROC of 4 combined models for predicting the risk of severe RSV-associated ALRTI in training and internal validation sets. ROC of AP (A), APA (B), ARA (C) and ARAP (D) were plotted to verify the diagnostic accuracy of these models in training set. ROC of AP (E), APA (F), ARA (G) and ARAP (H) were plotted in internal validation set Comparison of multiple models and selection of the best model Compared with AP and APA, ARA and ARAP demonstrated better calibrations, with non-significant Hosmer-Lemeshow test results in training set (AP: p < 0.001, APA: p = 0.022, ARA: p = 0.266, ARAP: p = 0.104) (Fig. 4 A). Comparable results were observed in the internal validation set (Fig. 4B). Additionally, the DCA of ARA and ARAP also showed greater net benefits than those of the AP and APA models in training and internal validation sets (Fig. 4C, D), especially at key thresholds 0.7 (Table S4). What’s more, NRI and IDI were used to evaluate whether incorporating prematurity, which was widely recognized as risk factor, into the ARA model would improve its predictive efficiency, the result showed no significant improvement (Table S5). Therefore, based on these comparative analyses and considering that prematurity is a widely accepted predictor of severe RSV-associated ALRTI, the ARAP was selected as the best model. ARAP was subsequently visualized as a nomogram (Fig. 4E). Fig. 4 Model calibration and decision curve analysis (DCA) of 4 combined models and ARAP nomogram. Calibration plots of 4 combined model in training (A) and internal validation sets (B) with Hosmer-Lemeshow test p-values, DCA of 4 combined model in training (C) and internal validation sets (D), ARAP nomogram for predicting the risk of severe RSV-associated ALRTI (E) Model calibration and decision curve analysis (DCA) of 4 combined models and ARAP nomogram. Calibration plots of 4 combined model in training (A) and internal validation sets (B) with Hosmer-Lemeshow test p-values, DCA of 4 combined model in training (C) and internal validation sets (D), ARAP nomogram for predicting the risk of severe RSV-associated ALRTI (E) The distribution of one patient was shown to illustrate how to use the nomogram. The red dot at each line represented the value of each of the four predictors for this patient. The total points were 188, corresponding to a probability of 0.904 for the risk of severe RSV-associated ALRTI. This patient was classified as high risk and recommended to receive intensive care. External validation and optimum application scope of ARAP ARAP was verified in an external validation cohort with an AUROC of 0.770 (95% CI: 0.677–0.863) (Fig. 5 A). To further optimize its clinical applicability, stratification analysis was conducted by gender and RSV type. In the gender-based analysis, ARAP achieved an AUROC of 0.855 (95% CI: 0.803–0.907) in the male subgroup and 0.767 (95% CI: 0.692–0.841) in the female subgroup (Fig. 5B, C). When stratified by RSV type [19], the model demonstrated an AUROC of 0.880 (95% CI: 0.809–0.951) in the RSV-A subgroup and 0.797 (95% CI: 0.744–0.849) in the RSV-B subgroup, respectively (Fig. 5D, E). Notably, ARAP achieved the highest AUROC value of 0.896 (95%CI 0.806–0.987) in the male RSV-A subgroup (Fig. 5 F), indicating the optimum application scope for ARAP. Fig. 5 ROC of ARAP for predicting the risk of severe RSV-associated ALRTI in stratification analysis ROC of the ARAP predicting the risk of severe RSV-associated ALRTI in external validating set (n = 136) (A), in male subgroup (n = 262) (B), in female subgroup (n = 202) (C), in RSV-A (n = 98) (D), in RSV-B (n = 366) (E), and in male-A subgroup (n=55) (F) ROC of ARAP for predicting the risk of severe RSV-associated ALRTI in stratification analysis ROC of the ARAP predicting the risk of severe RSV-associated ALRTI in external validating set (n = 136) (A), in male subgroup (n = 262) (B), in female subgroup (n = 202) (C), in RSV-A (n = 98) (D), in RSV-B (n = 366) (E), and in male-A subgroup (n=55) (F) Discussion In this study, we developed and externally validated a predictive model (ARAP) using four routinely available clinical variables (including age, RBC, ALT, and prematurity) to estimate the risk of severe RSV-associated ALRTI in pediatric patients. The ARAP model demonstrated good discrimination and calibration, with AUROC values of 0.828 in training set and 0.770 in external validation cohort. Given its simplicity, objectivity, and favorable performance, ARAP can help clinicians make early diagnosis, timely intervention and ultimately improve the prognosis of severe RSV-associated ALRTI patients. Recently, a predictive model was developed to predict severe RSV-associated bronchiolitis based on single-center data, including prematurity, weight on admission, lymphocyte percentage and outpatient use of glucocorticoids. This model can achieve an AUROC of 0.784 in the training set [13]. The ARAP model in our study was developed and further validated in multi-center data, with similar AUROC value of 0.828 and 0.770 in training and external validation cohorts. Liu et al. [14] proposed a more complex model comprising nine parameters, including age, prematurity, comorbidities, seizures, NLR, IL-6, LDH, D-dimer, and co-infection to predict severe RSV-associated ALRTI. This model can reach an AUROC of 0.865. Nevertheless, our ARAP model can also achieve comparable performance (AUROC = 0.828) with fewer variables, offering superior clinical applicability due to its simplicity and ease of use in real-world settings. In our study, three variables, including age, RBC and ALT, were identified as independent predictors and incorporated into the ARAP model. Among these, age was confirmed as an independent predictor of severe RSV-associated ALRTI by multivariate analysis (OR = 0.963, p = 0.064), which was consistent with previous research [9, 20, 21]. This result may be explained by age-related immune immaturity. In detail, young infants, particularly those under six months of age, have immature and naive immune systems and exhibit diminished activation of innate immune responses and inflammatory gene expression in response to RSV infection [22–24]. Furthermore, it was reported that maternal immunity, which young infants rely on heavily for protection against RSV, is limited and may contribute to their increased susceptibility to infection [25]. Specifically, inadequate activation of STAT3, a key anti-apoptotic factor, facilitated viral replication and extensive epithelial cell apoptosis in airway [26]. RBC count was identified to be correlated with severe RSV-associated ALRTI. Anemia has been identified as a risk factor for mortality in pediatric patients with severe community-acquired pneumonia [27]. Although rarely mentioned in previous RSV studies, blood transcriptome analysis identified high expression of erythroid cell-related phenotypes in severe RSV-associated ALRTI patients, potentially offering insight into the observed reduction in RBC count [28]. Anemia of inflammation may be also concerned in RSV-associated ALRTI, which is a kind of inflammatory condition. The proinflammatory cytokines induced by inflammation, promotes iron sequestration in macrophages, and ultimately leads to anemia [29, 30]. Additionally, these cytokines can damage renal epithelial cells and decrease erythropoietin production, thereby further reducing RBC count [29, 31]. Elevated ALT level was also identified as an independent predictor. It was reported that RSV could be isolated from infant liver tissue, and hepatic steatosis was observed in a fatal case of Reye’s syndrome associated with RSV infection, indicating an association between RSV infection and hepatic involvement [32, 33]. Furthermore, several studies had shown that elevated ALT level is related with increased severity of RSV-associated ALRTI [34, 35]. It can be explained that high viral load in severe RSV-associated ALRTI may cause viral spillover into the systemic circulation, resulting in hepatitis as an extrapulmonary manifestation [36]. Apart from direct RSV invasion of the liver, cytotoxic CD8⁺ T lymphocytes, which was independent from the virus, have recently been reported to mediate collateral hepatic injury [37]. Evidence indicated that a history of prematurity increases the risk of severe outcomes, such as respiratory failure and death, in children under two years of age with severe RSV-associated ALRTI [9, 38]. Prematurity disrupts terminal stages of lung development and maternal antibody transfer during late gestation, resulting in reduced lung volume and impaired pulmonary function, which may partly explain the increased susceptibility to severe RSV-associated ALRTI and adverse clinical outcomes [39, 40]. Unfortunately, prematurity was not selected as an independent predictor in our study. Collinearity diagnostics showed no collinearity between prematurity and other variables in current cohort (Table S2), indicating that its exclusion may not be linked to collinearity. Consistent with previous report [41], prematurity was also identified as a significant predictor in the univariate analysis of our study (OR = 2.785, p = 0.025). When prematurity was included in the final ARAP model, it demonstrated a slightly higher AUC compared with the ARA model (0.828 vs. 0.826 in the training set; 0.784 vs. 0.778 in the internal validation set; detailed in Fig. 3) and higher sensitivity (0.789 vs. 0.778 in the training set, Table S3). These improvements may enhance the prediction of severe RSV-associated ALRTI, supporting the role of prematurity as an important predictor. The limited number of premature infants (36/464, 7.76%) may have reduced its impact on prediction and contributed to its exclusion as an independent predictor. When categorized as moderate-to-late (defined as 32–37 weeks, n = 25) or very-to-extremely preterm (defined as < 32 weeks, n = 11), both groups failed to achieve clinically acceptable predictive performance for severe RSV-associated ALRTI, with AUC values of 0.515 (95% CI: 0.489–0.540) and 0.531 (95% CI: 0.509–0.554), respectively. More premature infants would be encouraged in future studies to better evaluate the predictive value of prematurity. Despite the addition of prematurity failed to show significant net benefit by NRI and IDI analysis in this study (Table S5), further validation in larger cohorts is needed. Key demographic and clinical characteristics between the training and external validation sets were compared and summarized them in Table S1. Compared with the training set, patients from the external validation cohort had a slightly lower proportion with a history of prematurity, higher proportion of breastfeeding and wheezing, and higher ALT level. Among the variables differing between these two sets, variations in the proportion of prematurity and breastfeeding reflected population composition differences. Additional, differences in the proportion of wheezing maybe influenced by the subjective physical examination, while differences in ALT were attributable to examination subjectivity and measurement bias. These heterogeneities between training set and external validation cohort may have contributed to the AUC decrease from 0.828 to 0.770. Age, prematurity, RBC count and ALT level, routinely available demographic and laboratory indices, were incorporated into the ARAP nomogram. Clinicians can use these values to calculate a total score corresponding to the risk of severe RSV-associated ALRTI. Application of ARAP is recommended as follows: (i) diagnose RSV-associated ALRTI according to the diagnostic criteria; age, prematurity, RBC count and serum ALT level are also measured on admission; (ii) the total points are calculated by summing the points identified on the “Point” scale for each predictor. By comparing the “Total points” scale and the “Risk” scale, the individual risk level of severe RSV-associated ALRTI could be obtained (Fig. 4E); (iii) patients with a risk probability greater than 0.7 are classified as high risk and should receive intensive care, while others may proceed with standard management. ARAP provides clinicians with a convenient clinical decision-making support tool for assessing the risk of severe RSV-associated ALRTI. Using stratification analysis, the male RSV-A subgroup was identified as the optimum application scope of ARAP with the highest AUROC of 0.896 (Fig. 5 F). Despite slightly lower AUROC values in female (0.767, 95% CI: 0.692–0.841) and RSV-B (0.797, 95% CI: 0.744–0.849) subgroups, the model still maintained moderate predictive accuracy. Additionally, in training set, gender and RSV type distributions showed no statistically difference in males (61.1% vs.52.4%, p = 0.205) or RSV-A (22.2% vs. 26.7%, p = 0.417) between the severe and non-severe RSV-associated ALRTI groups (Table 1). This may indicate that the optimal application is not up to the proportion of gender or RSV A/B type in our cohort, but to host’s immunological status or the pathogen’s biological characteristics, warranting further investigation in future studies. This study has several limitations. Although 600 participants from two medical centers were included, larger prospective, multicenter studies are needed to further validate the ARAP model. Additionally, all patients in this study were enrolled from general ward, resulting in a limited number of comorbid patients in this study. This small size hindered their effective selection as predictors and retention in the model. Moreover, as the model was derived exclusively from an Asian pediatric population, its generalizability to other ethnic groups remains uncertain. Further application in diverse patient populations and in centers outside China is strongly recommended to confirm its broader applicability. Conclusion The ARAP model was identified as a simple and effective predictive model of assessing the risk of severe RSV-associated ALRTI in pediatric patients, especially in the male RSV-A subgroup. Supplementary Information Supplementary Material 1 Supplementary Material 1