Abstract
The introduction of pertussis vaccination in the United States (US) in the 1940s has greatly reduced its burden. However, the incidence of pertussis is difficult to quantify, as many cases are not laboratory-confirmed or reported, particularly in adults. This study estimated pertussis incidence in a commercially insured US population aged <50 years. Data were extracted from IMS' PharMetrics Plus claims database for patients with a diagnosis of pertussis or cough illness using International Classification of Diseases (ICD-9) codes, a commercial outpatient laboratory database for patients with a pertussis laboratory test, and the Centers for Disease Control influenza surveillance database. US national pertussis incidence was projected using 3 methods: (1) diagnosed pertussis, defined as a claim for pertussis (ICD-9 033.0, 033.9, 484.3) during 2008-2013; (2) based on proxy pertussis predictive logistic regression models; (3) using the fraction of cough illness (ICD-9 033.0, 033.9, 484.3, 786.2, 466.0, 466.1, 487.1) attributed to laboratory-confirmed pertussis, estimated by time series linear regression models. Method 1 gave a projected annual incidence of diagnosed pertussis of 9/100,000, which was highest in those aged <1 year. Method 2 gave an average annual projected incidence of 21/100,000. Method 3 gave an overall regression-estimated weighted annual incidence of pertussis of 649/100,000, approximately 58-93 times higher than method 1 depending on the year. These estimations, which are consistent with considerable underreporting of pertussis in people aged <50 years and provide further evidence that the majority of cases go undetected, especially with increasing age, may aid in the development of public health programs to reduce pertussis burden.