Table 1.

Input parameters for the risk model to estimate true incidence of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS)

Steps in estimation	Parameters	S*	N†	Distribution‡	Median	95% CrI	Source
1. HUS notifications	Incidence of notified cases	260	327 × 108§	Gamma(260, 3 × 10−9)	8 × 10−7	7 × 10−7–9 × 10−7	German notification data
(a) Adjustment for underreporting separately for cases treated in PNCs and non-PNCs	Proportion of HUS-notifications treated by PNCs	153	254	Beta(154, 102)	0·60	0·54–0·66	National active and passive surveillance, unpublished
	Completeness of HUS-notification from PNCs	153	183	Beta(154, 31)	0·83	0·78–0·88	National active and passive surveillance, unpublished
	Multiplication factor to extrapolate completeness of notification from PNCs to non-PNCs	−	−	Pert(0·1, 0·5, 1)	0·51	0·21–0·84	Assumption
(b) Proportion of STEC-associated HUS in enteropathic HUS cases	Proportion of STEC-associated-HUS	327	394	Beta(328, 68)	0·83	0·79–0·86	Gerber et al. 2002 [16]
(c) Proportion of O157 and non-O157 in STEC-associated HUS	Proportion of O157 in STEC-associated HUS	138	207	Beta(494, 239)	0·67	0·64–0·71	Gerber et al. 2002 [16]
		355	524				Mellmann et al. 2008 [17]
(d) Number of laboratory-confirmed STEC-GE cases per HUS case	Proportion HUS in laboratory-confirmed O157 STEC	3	27	Beta(4, 25)	0·13	0·04–0·28	Werber et al. 2007 [18]
	Proportion HUS in laboratory-confirmed non-O157 STEC	2	149	Beta(3, 148)	0·02	0·00–0·05	Werber et al. 2007 [18]
(e) Proportion of bloody diarrhoea in O157 and non-O157 in STEC-GE cases	Proportion of cases experiencing bloody O157-associated diarrhoea	10	27	Beta(11, 18)	0·38	0·22–0·56	Werber et al. 2007 [18]
	Proportion of cases experiencing bloody non-O157-associated diarrhoea	16	149	Beta(17, 134)	0·11	0·07–0·17	Werber et al. 2007 [18]
(f) Underascertainment of bloody and non-bloody diarrhoea	Proportion of patients visiting physicians with bloody diarrhoea	21	41	Beta(22, 21)	0·51	0·36–0·66	Haagsma et al. 2013 [19]
	Proportion of patients visiting physicians with non-bloody diarrhoea	458	1342	Beta(555, 1093)	0·34	0·31–0·36	Haagsma et al. 2013 [19]
		96	304				Hauri et al. 2011 [20]
	Proportion of physicians taking laboratory samples from patients with bloody diarrhoea	10	20	Beta(11, 11)	0·50	0·30–0·70	Haagsma et al. 2013 [19]
	Proportion of physicians taking laboratory samples from patients with non-bloody diarrhoea	155	456	Beta(170, 383)	0·31	0·27–0·35	Haagsma et al. 2013 [19]
		14	95				Hauri et al. 2011 [20]
	Proportion of stool samples tested for STEC from patients with bloody diarrhoea	−	−	None	1·00	1·00–1·00	Kist et al. 2013 [21]
	Proportion of stool samples tested for STEC from patients with non-bloody diarrhoea	−	−	Pert(0·1, 0·8, 1)	0·74	0·37–0·96	Kist et al. 2013 [21], assumption

CrI, Credible interval; PNC, paediatric nephrology centre.

^*Nominator.

^†Denominator.

^§The unit of measurement is person-years-at-risk for this parameter.

^‡For Gamma(r, λ), r = s and λ = 1/N; for Beta(a, b), a = Sum(s) + 1 and b = Sum(N) – Sum(s) + 1.