Table 3.

Ratios between RfDs based on the BMDL₁₀ and NOAEL, and RfDs based on the SNCD, expressed as medians and lower 5th (P05) and upper 95th (P95) percentiles across the n data sets considered.

RfD ratioa	Caseb	n	Median	P05	P95
RfDBMDL10: RfDSNCD1.0		1		439		1.0		0.50		1.6
		2		106		0.98		0.51		1.8
		3		217		0.89		0.39		1.7
		4		222		1.0		0.75		1.5
RfDBMDL10: RfDSNCD0.67		1		665		1.0		0.66		1.3
		2		124		1.0		0.66		1.1
		3		243		0.93		0.42		1.6
		4		422		1.1		0.84		1.1
RfDNOAEL: RfDSNCD1.0		1		439		1.5		0.41		7.6
		2		106		1.4		0.41		8.6
		3		217		2.4		0.65		9.3
		4		222		1.0		0.38		3.2
RfDNOAEL: RfDSNCD0.67		1		665		1.2		0.39		7.7
		2		124		1.2		0.40		9.9
		3		243		2.2		0.42		11
		4		422		1.0		0.39		3.3
aThe RfDs based on the BMDL10 and NOAEL were established by dividing the respective PoD by a UF of 100; the RfD based on the SNCD was established by linear extrapolation from the UERSNCD,and it corresponds to a target extra risk of 1/1,000. bDifferent subsets of the 786 data sets were considered, as represented by the following cases: Case 1: All data sets adequate for dose–response modeling [according to the criteria in Supplemental Material, Section 1 (http://dx.doi.org/10.1289/ehp.1003327)], and for which an SNCD was derived. Case 2: Data sets with the lowest NOAEL in each selected technical report included in case 1. Case 3: Data sets in case 1 including at most one dose for which the observed extra risk was > 0 and ≤ 0.2; the observed extra risk at a given dose, i, was calculated as [xi/ni – p(0)]/[1 – p(0)], where p(0) is the point estimate of the background risk. Case 4: Data sets in case 1 including more than one dose for which the observed extra risk was > 0 and ≤ 0.2.