. 2012 Mar 7;279(1739):2736–2743. doi: 10.1098/rspb.2011.2464

Table 1.

Predicted lag times associated with the maximum correlation (l_max; in weeks) for models- representing hypotheses regarding the aetiology of Kawasaki disease.

model	mixing	duration of infectiousness
model	mixing	one week	four weeks	eight weeks	16 weeks
SIR	homogenous	−1^a	−1	−1	0
	assortative	10	15	20	−24^c
	self-reported	8	11	12	11
	RAS	−12^bc	−12^bc	−12^bc	−13^bc
SEIR	homogenous	−1^a	1	1	1
	assortative	12	15	19	26^c
	self-reported	11	13	15	15
	RAS	−12^bc	−11^bc	−12^bc	−13^bc
SIRS	homogenous	−1	−1	−1	−1
	assortative	−1	0	−4	−12^c
	self-reported	−11^c	−11^c	−11^c	−11^c
	RAS	−13^bc	−12^bc	−11^bc	−11^bc
co-infection	homogenous	−2^d	3^a	20	−25^c
	assortative	10^a	15^a	18	23^c
	self-reported	9^d	19^a	16	22^c
	RAS	−11^bc	−13^bc	−17^bc	−21^bc

^aEpidemics occurred biennially.

^bAge distribution of cases predicted under realistic age-structured (RAS) mixing is inconsistent with the age distribution of Kawasaki disease hospitalizations.

^cPredicted lag time within ±12 weeks of the observed lag time for Kawasaki disease.

^dEpidemics occurred triennially.