2636. Distinguishing Pertussis from Viral Mimickers: Development and Validation of a Clinical Prediction Score

Abstract

BackgroundPertussis is often confused with respiratory viral infections, leading to misdiagnosis and overuse of antibiotics. Distinguishing the two entities more accurately can help optimize care.MethodsWe reviewed the charts of children under 18 years of age who presented to Sultan Qaboos University Hospital in Muscat, Oman and were tested for Bordetella pertussis by PCR between 2013 and 2018 (discovery cohort). Clinical and laboratory data were collected from the electronic patient record and analyzed. Backward conditional logistic regression was used to identify independent predictors of laboratory-confirmed pertussis cases. The Muscat Pertussis Index (MPI) score was developed based on the logistic regression model. The MPI score was retrospectively validated on a separate cohort of pediatric patients who presented to the Royal Hospital- Oman’s largest pediatric center- between 2017 and 2018, and were similarly tested for pertussis (validation cohort). Ethical approval of the study was obtained formally for both sites.Results354 patients were enrolled in the discovery cohort. 196 (55%) were male, and the median age was 10 weeks (IQR, 6–16). 57 (16%) patients tested positive for B. pertussis by PCR, while 266 (75%) tested positive for respiratory viruses. 32 (9%) patients had both pertussis and a viral co-infection and 63 (18%) were negative for both. 255 (72%) patients received macrolide antibiotics. Younger age, fewer vaccine doses, contact with a sick adult, longer symptom duration, paroxysmal cough, cyanosis, post-tussive emesis, apnea, lymphocytosis and thrombocytosis were significantly associated with pertussis (Table 1). After logistic regression, independent predictors of pertussis were longer symptom duration, lymphocytosis, paroxysmal cough, lack of fever, cyanosis and age under 8 weeks. This formed the basis for creating the MPI score (Table 2). The MPI score was validated on a cohort of 122 patients. Higher MPI scores correlated significantly with confirmed pertussis cases (area under the receiver operating characteristics curve = 0.899, P < 0.001, Figure 1 and Table 3).ConclusionThe majority of suspected pertussis cases were actually due to viral mimickers. The MPI score can predict likely cases of pertussis before laboratory confirmation. Future validation in more diverse settings would help expand its applicability. Disclosures All authors: No reported disclosures.