Prior candidate gene studies that examined OPRM1 haplotypes with rs1799971 suggested that other variants may explain its equivocal association with OA39,40. Subject level genotypes for haplotype analysis were available from a subset of cohorts contributing to the gSEM analysis (Fig. 2). In this subset of cohorts, the single variant results for rs9478500 (beta = 0.205, p = 2.43 × 10–9) were strong, but those for rs1799971 were weak (beta = − 0.058, p = 0.135). Comparison of the three haplotypes formed by rs1799971, rs9478500, and the other GWS OPRM1 variants (Fig. 2a) further weakened evidence for an association with OA being driven by rs1799971 (Fig. 2b comparison 1: p = 0.52, beta = − 0.026, SE = 0.0397) and strengthened evidence for an association with OA being driven by the effect of the non-rs1799971 variants (comparison 2: p = 1.63 × 10–10, beta = 0.2303, SE = 0.036, Fig. 2b).
