Risk factors and prediction model of severe pertussis in infants < 12 months of age in Tianjin, China

BACKGROUND: To identify risk factors associated with the prognosis of pertussis in infants (< 12 months). METHODS: A retrospective study on infants hospitalized with pertussis January 2017 to June 2019. The infants were divided into two groups according to the severity of disease: severe pertussis and non-severe pertussis groups. We collected all case data from medical records including socio-demographics, clinical manifestations, and auxiliary examinations. Univariate analysis and Logistic regression were used. RESULTS: Finally, a total of 84 infants with severe pertussis and 586 infants with non-severe pertussis were admitted. The data of 75% of the cases (severe pertussis group, n = 63; non-severe pertussis group, n = 189) were randomly selected for univariate and multivariate logistic regression analysis. The results showed rural area [P = 0.002, OR = 6.831, 95% CI (2.013–23.175)], hospital stay (days) [P = 0.002, OR = 1.304, 95% CI (1.107–1.536)], fever [P = 0.040, OR = 2.965, 95% CI (1.050–8.375)], cyanosis [P = 0.008, OR = 3.799, 95% CI (1.419–10.174)], pulmonary rales [P = 0.021, OR = 4.022, 95% CI (1.228–13.168)], breathing heavily [P = 0.001, OR = 58.811, 95% CI (5.503–628.507)] and abnormal liver function [P < 0.001, OR = 9.164, 95% CI (2.840–29.565)] were independent risk factors, and higher birth weight [P = 0.006, OR = 0.380, 95% CI (0.191–0.755)] was protective factor for severe pertussis in infants. The sensitivity and specificity of logistic regression model for remaining 25% data of severe group and common group were 76.2% and 81.0%, respectively, and the consistency rate was 79.8%. CONCLUSIONS: The findings indicated risk factor prediction models may be useful for the early identification of severe pertussis in infants.