Predicting pertussis in infants

Abstract

Background: The incidence of reported cases of pertussis is increasing, despite high rates of vaccination among infants and children. The burden of disease, and rates of complication and death are highest among infants. The limited availability of a timely reliable confirmatory test for pertussis hinders early identification of infected infants.

Objective: Our objective is to identify clinical and laboratory predictors for pertussis among infants.

Methods: Demographic, clinical, and laboratory data were collected from the medical records of all infants aged 12 months or younger who underwent confirmatory testing (culture, direct fluorescent assay, or polymerase chain reaction) for Bordetella pertussis from January 1, 2001, to July 31, 2005. The association of 15 variables with a positive pertussis test was analyzed using univariate and multivariate analysis, and when appropriate, using receiver operating characteristics.

Results: We reviewed the medical records of 141 infants who were tested for pertussis. The mean age was 88 days, and the most common chief complaints were cough and breathing difficulty. Eighteen patients (13%) had a positive pertussis test, and 123 (87%) had a negative test. Bronchiolitis and upper respiratory tract infection were the most common discharge diagnoses among infants with a negative test. The 2 groups were similar with respect to sex, history of cough, vomiting, fever, symptoms of apparent life-threatening event, presence of fever, or hypoxia, and heart rate. Infants who were younger (55 days vs 93 days, P = 0.02), evaluated between July and October (23% vs 9%, P = 0.02), less tachypneic (39 breaths/min vs 47 breaths/min, P = 0.02), had higher white blood cell counts (20,000/microL vs 15,000/microL, P = 0.02), higher percentage of lymphocytes (72 vs 55, P = 0.00), and higher absolute lymphocyte counts ([ALC] 14,536/microL vs 8357/microL, P = 0.00) were more likely to have a positive test. Receiver operating characteristics for ALC demonstrated an area under the curve of 0.81, with a 95% confidence interval of 0.72 to 0.90. An ALC cutoff point of 9400 was determined to maximize sensitivity (89%) and specificity (75%). The negative predictive value of this cutoff point was 97%, and the positive likelihood ratio was 3.6, with a 95% confidence interval of 2.3 to 5.4.

Conclusions: Among infants who underwent confirmatory testing for pertussis, those who are younger, evaluated between July and October, less tachypneic, have higher white blood cell counts, higher percentage of lymphocytes, and higher ALCs are more likely to have a positive test. The ALC was the best predictor of pertussis, and an ALC of less than 9400/microL excluded almost all infants without pertussis.