About those assumptions: The problem with models

This is the stuff that friend-of-the-barfblog Don Schaffner gets excited about: Modelling.

Belgian researchers attempted to predict future rates of specific foodborne illnesses to guide policy efforts and came up with this:

Salmonellosis, campylobacteriosis and listeriosis are foodborne diseases. We estimated and forecasted the number of cases of these three diseases in Belgium from 2012 to 2020, and calculated the corresponding number of disability-adjusted life years (DALYs). The salmonellosis time series was fitted with a Bai and Perron two-breakpoint model, while a dynamic linear model was used for campylobacteriosis and a Poisson autoregressive model for listeriosis.

The average monthly number of cases of salmonellosis was 264 (standard deviation (SD): 86) in 2012 and predicted to be 212 (SD: 87) in 2020; campylobacteriosis case numbers were 633 (SD: 81) and 1,081 (SD: 311); listeriosis case numbers were 5 (SD: 2) in 2012 and 6 (SD: 3) in 2014.

After applying correction factors, the estimated DALYs for salmonellosis were 102 (95% uncertainty interval (UI): 8–376) in 2012 and predicted to be 82 (95% UI: 6–310) in 2020; campylobacteriosis DALYs were 1,019 (95% UI: 137–3,181) and 1,736 (95% UI: 178–5,874); listeriosis DALYs were 208 (95% UI: 192–226) in 2012 and 252 (95% UI: 200–307) in 2014.

New actions are needed to reduce the risk of foodborne infection with Campylobacter spp. because campylobacteriosis incidence may almost double through 2020.

Burden of salmonellosis, campylobacteriosis, and listeriosis: a time series analysis



Maertens de Noordhout et al.