Spatiotemporal patterns in pertussis incidence — United States, 2000–2017

  • Public health surveillance
Export to CSV
Background:
Pertussis, or whooping cough, is a highly contagious, vaccine-preventable respiratory disease. Historically, pertussis incidence was cyclic with peaks in disease every 3-5 years. In the United States, reported pertussis has increased over the past few decades despite high vaccination coverage; however, there is no clear national spatiotemporal pattern. We aimed to assess: 1) the spatiotemporal distribution of pertussis in the United States, and 2) whether geographically distinct areas share similar temporal patterns.

Methods:
We used pertussis cases reported through the National Notifiable Diseases Surveillance System, and county population estimates from the U.S. Census Bureau, for 2000-2017. County-level case counts were aggregated by month. To assess the distribution of pertussis cases, and identify spatiotemporal clusters during our study period, we used Kulldorf’s spatiotemporal scan statistic (p<0.01). For each cluster of identified counties, wavelet analysis was used to quantify the timing and periodicity of pertussis incidence.

Results:
National pertussis incidence over 2000-2017 averaged 6.4 cases/100,000 annually with peaks in 2004-2005, 2010, 2012, and 2014. We detected spatiotemporal clusters of high pertussis incidence, with the geographically largest clusters in the East North Central (relative risk (RR) of disease within cluster compared to outside: 3.9), New England (RR: 3.1), and northern Mountain regions (RR: 2.4). On average, clusters spanned 24 (1-331) counties and lasted 28 (4-108) months. Although there was substantial variability in the temporal pattern for each cluster, spatially distinct areas can be grouped by similar dominant periods of 12 months or >20 months.

Conclusion:
Pertussis has disproportionately affected certain areas across the nation. A better understanding of the current spatiotemporal patterns of pertussis across the United States will allow us to better characterize current epidemiology, potentially helping predict and plan for the occurrence of future outbreaks.

Please abstracts [at] tephinet [dot] org (email us) if you have any corrections.

If this abstract has been converted into a full article, please abstracts [at] tephinet [dot] org (email us) the link. We would love to help promote your work.