Risk factors and prediction score model for unplanned readmission among neonates with NRDS under one year of age: A retrospective cohort study

OBJECTIVE: This study aimed to analyze the risk factors and establish a prediction score model for unplanned readmission among neonates with neonatal respiratory distress syndrome (NRDS) for respiratory problems under one year of age. METHODS: This retrospective cohort study enrolled 230 neonates with NRDS who were admitted between January 2020 and December 2020. The infants were classified into two subgroups based on whether they were readmitted for respiratory problems under one year of age: readmit group and non-readmit group. Readmission risk factors for NRDS were analyzed by logistic regression and a prediction score model was generated. RESULTS: Among the 230 enrolled infants, 51 (22%) were readmitted, and 179 (78%) were not readmitted. In univariate analysis, compared with non-readmit group infants, readmit group infants had a significantly younger birth gestational age (31.9 ± 2.3 vs. 32.8 ± 2.5 weeks, p = 0.012), lower birth weight (1,713.7 ± 501.3 g vs. 1,946.8 ± 634.4 g, p = 0.007), older age at discharge (41.7 vs. 31.7 days, p = 0.012), higher proportion of necrotizing enterocolitis (NEC) (31% vs. 16%, p = 0.016), higher rate of blood transfusion (39% vs. 25%, p = 0.049), higher rate of postnatal dexamethasone (DEX) administration (28% vs. 9.5%, p = 0.001), and higher rate of home oxygen therapy (HOT) (57% vs. 34%, p = 0.003). Moreover, readmit group infants had significantly longer antibiotic days usage (12.0 vs. 10.0 days, p = 0.026) and a longer duration of hospital stay (41.0 vs. 31.0 days, p = 0.012) than non-readmit group infants. The multivariate logistic regression analysis showed that taking readmission as a target variable, postnatal DEX administration (OR: 2.689, 95% CI: 1.168–6.189, p = 0.020), HOT (OR: 2.071, 95% CI: 1.060–4.046, p = 0.033), and NEC (OR: 2.088, 95% CI: 0.995–4.380, p = 0.051) could be regarded as risk factors for readmission. A scoring model predicting readmission was administered with a positive predictive value of 0.651 (95% CI: 0.557–0.745, p = 0.002), with a sensitivity of 0.412 and a specificity of 0.888 at a cut-off of 3.5 points, which were evaluated on the receiver operating characteristic curve. CONCLUSIONS: Postnatal DEX administration, HOT, and NEC were risk factors for readmission of NRDS. NRDS infants with a predictive score of 3.5 points or more were at high risk for unplanned readmission.