Since, as an exploratory study, we did not correct for multiple comparisons, definitive interpretation of these findings will require confirmation in larger cohort studies. Nevertheless, our findings support multiple pathways being involved in host response to IPD. Recent studies also suggest that additional genes in the toll-like receptor-signaling pathway may influence response to IPD. Furthermore, the collectin MBL2 had variants overrepresented in pneumococcal bacteremia and meningitis, but not for overall IPD. This suggests the possibility of syndrome-specific host genetic associations, but our study was underpowered to definitively evaluate this. For this analysis, we took an indirect association approach by selecting and genotyping SNPs that are either causative SNPs or in LD with the causative SNP. The latter situation most likely applies to the majority of SNPs found associated with IPD in this study, as 18 of the 27 associated SNPs are located in introns. Furthermore, of the four SFTPD variants associated with IPD in either EA and/or AA , all are intronic. Notably, in AMN107 Src-bcr-Abl inhibitor SeattleSNPs EA, rs17886286 and rs1998374 are in complete or high LD with rs3088308, a coding INCB18424 JAK inhibitor non-synonymous SNP, while in SeattleSNPs AA, these SNPs have little to no LD with rs3088308 and are not associated with IPD in AA. The two SFTPD variants that are associated with IPD in AA are in moderate LD with a different coding non-synonymous SNP, rs4469829, which is monomorphic in EA. Furthermore, SFTPD variant rs721917, a non-synonymous SNP known to reduce serum levels of SP-D in EA, is in LD with rs1998374 in AA but not EA. Thus, differences in LD patterns between ancestral populations may help to explain the disparate signals observed in EA compared to AA. Our primary goal was to assess the feasibility of cross-linking surveillance data with an nDBS repository to perform tagSNP genomic studies, and toward this end we were highly successful: 82% of surveillance cases were linked to an nDBS, and 88% of samples successfully genotyped. Several key issues associated with this experience deserve emphasis. First, the completeness of IPD case surveillance in ABCs through use of active surveillance methods and routine audits of laboratory records combined with the overall low incidence of IPD in the general population support our assumption that controls were at low risk of having had IPD outside the surveillance time-period. Second, efficient linking of surveillance cases to nDBS samples was critical to minimize bias, but this linkage depends on consent requirements for nDBS use, which differ by state and continue to evolve.