Our analysis of IPF suggests that a paradoxical association of the strong risk locus MUC5B with increased survival may be due to index event bias, and that in fact this gene may well cause decreased survival. It has been hypothesised that carriers of MUC5B risk alleles experience a milder form of disease, in line with the clinical heterogeneity of IPF27. While associations with prognosis can be explained by disease heterogeneity, they remain susceptible to index event bias whether or not disease subtype is accounted for (Fig. 4). It is therefore important to account for the bias to inform the interpretation of genetic associations with prognosis. Here, the reversal of direction for the MUC5B survival effect is largely due to its exceptionally high odds ratio for susceptibility. However, our result, while significant, is imprecise and based on a sample that is small by current standards. It is crucial to replicate this result in larger samples or meta-analyses.
