Abstract:

Objective To investigate the prevalence and symptoms of pertussis in patients with cough, and to predict individual risk based on the combination of symptoms. Methods A total of 1 025 patients with cough or contact with pertussis patients were included. Pertussis was confirmed by nasopharyngeal swab PCR. Patients were divided into the juvenile group (278 cases) and the adult group (747 cases) according to age. The duration of cough from onset to study participation, the visual analogue (VAS) score of cough degree and the number of basic symptoms (paroxysmal cough, vomiting after cough, crowing cough, pauses in breathing after cough and fever) were compared between the two groups. The confirmed patients were further grouped by age, and the different symptoms were compared. 70% of the sample was used as the training set. Based on the combination of symptoms (paroxysmal cough, post-cough vomiting, chick-crooning cough and pauses in inspirations after coughing), multivariate Logistic regression was used to establish the prediction model and draw the nomogram. 30% of the sample was used as the validation set, and the receiver operating characteristic (ROC) curve was drawn. The differentiation of the area under the curve (AUC) evaluation model was calculated. The calibration degree of the model was evaluated by Hosmer-Lemeshow test, and calibration curve was drawn to evaluate the model. Results By PCR, 163 cases (15.9%) were confirmed as pertussis. The juvenile group had a longer duration of cough from onset to study participation than the adult group (P < 0.05). The VAS score of cough severity was higher, and the number of basic symptoms of pertussis was more (P < 0.01). In confirmed cases, the proportion of paroxysmal cough, vomiting after cough, crowing cough and inspiratory pause after cough was higher in the juvenile group than that in the adult group (P < 0.01). In the diagnosed cases, the incidence of paroxysmal cough and post-cough vomiting were higher in the <1-year-old group compared to the 1-9-year-old group and the ≥10-year-old group (P < 0.05). The combination of paroxysmal cough, vomiting after cough, crowing cough and inspiratory pause after cough was selected by Logistic regression analysis to establish a nomogram model. The AUC of this model in the training set was 0.852, and the Hosmer-Lemeshow test χ²=0.208, P = 0.901, and in the verification set, the AUC was 0.899, and the Hosmer-Lemeshow test χ²=4.202, P = 0.122. The predicted value in the calibration curve was very close to the theoretical value in the training set and the verification set, and the fitting degree was high. Conclusion The infection rate of pertussis is high in patients with cough. The nomogram model based on combined symptoms has a better prediction effect on pertussis differentiation, which can provide reference for the monitoring of pertussis.

Key words: whooping cough, cough, forecasting, nomograms, polymerase chain reaction