Rxivist logo

Slope-Hunter: A robust method for index-event bias correction in genome-wide association studies of subsequent traits

By Osama Mahmoud, Frank Dudbridge, George Davey Smith, Marcus Munafò, Kate Tilling

Posted 31 Jan 2020
bioRxiv DOI: 10.1101/2020.01.31.928077

Background: Studying genetic associations with prognosis (e.g. survival, disability, subsequent disease events) is problematic due to selection bias - also termed index event bias or collider bias - whereby selection on disease status can induce associations between causes of incidence with prognosis. A current method for adjusting genetic associations for this bias assumes there is no genetic correlation between incidence and prognosis, which may not be a plausible assumption. Methods: We propose an alternative, the 'Slope-Hunter' approach, which is unbiased even when there is genetic correlation between incidence and prognosis. Our approach has two stages. First, we use cluster-based techniques to identify: variants affecting neither incidence nor prognosis (these should not suffer bias and only a random sub-sample of them are retained in the analysis); variants affecting prognosis only (excluded from the analysis). Second, we fit a cluster-based model to identify the class of variants only affecting incidence, and use this class to estimate the adjustment factor. Results: Simulation studies showed that the Slope-Hunter method reduces type-1 error by between 49%-85%, increases power by 1%-36%, reduces bias by 17%-47% compared to other methods in the presence of genetic correlation and performs as well as previous methods when there is no genetic correlation. Slope-Hunter and the previous methods perform less well as the proportion of variation in incidence explained by genetic variants affecting only incidence decreases. Conclusions: The key assumption of Slope-Hunter is that the contribution of the set of genetic variants affecting incidence only to the heritability of incidence is at least as large as the contribution of those affecting both incidence and prognosis. When this assumption holds, our approach is unbiased in the presence of genetic correlation between incidence and progression, and performs no worse than alternative approaches even when there is no correlation. Bias-adjusting methods should be used to carry out causal analyses when conditioning on incidence.

Download data

  • Downloaded 269 times
  • Download rankings, all-time:
    • Site-wide: 54,780 out of 85,056
    • In genetics: 3,116 out of 4,463
  • Year to date:
    • Site-wide: 7,688 out of 85,056
  • Since beginning of last month:
    • Site-wide: 20,429 out of 85,056

Altmetric data

Downloads over time

Distribution of downloads per paper, site-wide


Sign up for the Rxivist weekly newsletter! (Click here for more details.)