Real-Time
Prediction of West Nile Virus Neuroinvasive Disease Case Counts
Accounting for Reporting Delays
|
Brad
Biggerstaff, Centers for Disease Control and Prevention
Monday, 03 November 2003
4:10 PM
E202 Engineering Building
ABSTRACT
West
Nile virus (WNV) epidemics have occurred in the US since it was first
recognized in the Western Hemisphere in an outbreak in Queens, New York, in
1999. The 2002 WNV epidemic was the largest WNV outbreak on record, and the
largest arbovirus outbreak recorded in the US. With the spread of the virus
westward across the US, there was every indication that a large outbreak
was again possible in 2003. Routine surveillance for WNV illness on the
national level has been in place since 2002, and epidemiologists with the
Centers for Disease Control and Prevention (CDC) regularly track the
epidemic using daily updated
WNV
case counts, reported to the CDC from state health departments. There are
naturally delays from the time a person takes ill from a WNV infection and
the time that such illness is reported to CDC. With the 2003 epidemic, we
augmented the existing surveillance case tally to account for such
reporting delays. The nonparametric methods of Lawless (1994) and
Kalbfleisch and Lawless (1989) were used to provide daily adjustments to
the case counts. The methodology and
application
are presented, along with some description of the reporting delay
distributions observed during the 2002 and 2003 epidemics. Additionally,
real-time comparison of the national 2003 outbreak with the one in 2002 is
made using the reporting delay-adjusted predictions. Finally, the utility
of the approach is discussed, and planned extensions to the current implementation
are outlined.
Joint
work with Alicia Johnson
Refreshments will be served at
3:45 p.m. in Room 008 of the Statistics Building
|
