How are incidence trends examined in FoodNet?
FoodNet uses a main-effects, log-linear Poisson regression model (negative binomial) to estimate the incidence of the pathogens and syndrome under surveillance. This model controls for the increase in the number of FoodNet sites and its surveillance population since 1996 and for variation in the incidence of infections among sites. The average annual incidence for the previous 3 years (2005-2007) and for 1996-1998 (1997-1998 for Cryptosporidium), the first years of surveillance, are used as comparison periods. The fitted model is used to calculate the estimated change in incidence (relative rate) between 2008 and the comparison periods, along with 95% confidence intervals (CI). In the percent change figures, the percent change for the current year is compared with the appropriate comparison period and graphed along with the 95% confidence interval. In the relative rate figures, the relative rate for each year is graphed.
Results:
By pathogen
Compared with 1996-1998 period
Percent change
Relative rates (graph 1)
Relative rates (graph 2)
Compared with 2005-2007 period
Percent change
By Salmonella serotype
Compared with 1996-1998 period
Percent change
Relative rates
Compared with 2005-2007 period
Percent change
|