Nature 414, 748–751 (2001).

There was an error in equations (1) in this Letter. The θI terms in dI/dt and dQ/dt should have been omitted. Thus, the set of differential equations should have been:

and the definition of θ in Table 2 should have read ‘the proportion of infectious individuals from En that are quarantined’. This change only affects the analysis of the data for Kosovo. The value of R0 that gives the best fit to the data for Kosovo is now 10 (5 discounting hospital-associated cases), rather than 10.8 (5.4 discounting hospital-?associated cases), as stated in the original paper in Table 1. This small error does not alter our conclusions.