Bayesian methods for meta-analysis of causal relationships estimated using genetic instrumental variables

Stephen Burgess; Simon G Thompson; Grahame Thompson; G Andrews; N J Samani; Anne Mette Hall-Andersen; P Whincup; R Morris; D A Lawlor; G Davey Smith; N Timpson; S Ebrahim; Y Ben-Shlomo; M Brown; S Ricketts; M Sandhu; A Reiner; B Psaty; L Lange; M Cushman; J Hung; P Thompson; J Beilby; N Warrington; L J Palmer; B G Nordestgaard; Anne Tybjærg-Hansen; J Zacho; C Wu; G Lowe; I Tzoulaki; Mahesha Kumari Perera; J F Yamamoto; B Chiodini; M Franzosi; G J Hankey; K Jamrozik; L Palmer; E Rimm; J Pai; S Heckbert; J Bis; S Anand; J Engert; R Collins; R Clarke; O Melander; G Berglund; P Ladenvall; L Johansson; CRP CHD Genetics Collaboration

doi:http://dx.doi.org/10.1002/sim.3843

Bayesian methods for meta-analysis of causal relationships estimated using genetic instrumental variables

Stephen Burgess, Simon G Thompson, Grahame Thompson, G Andrews, N J Samani, Anne Mette Hall-Andersen, P Whincup, R Morris, D A Lawlor, G Davey Smith, N Timpson, S Ebrahim, Y Ben-Shlomo, M Brown, S Ricketts, M Sandhu, A Reiner, B Psaty, L Lange, M CushmanJ Hung, P Thompson, J Beilby, N Warrington, L J Palmer, B G Nordestgaard, Anne Tybjærg-Hansen, J Zacho, C Wu, G Lowe, I Tzoulaki, Mahesha Kumari Perera, J F Yamamoto, B Chiodini, M Franzosi, G J Hankey, K Jamrozik, L Palmer, E Rimm, J Pai, S Heckbert, J Bis, S Anand, J Engert, R Collins, R Clarke, O Melander, G Berglund, P Ladenvall, L Johansson, CRP CHD Genetics Collaboration

21 Citations (Scopus)

Abstract

Genetic markers can be used as instrumental variables, in an analogous way to randomization in a clinical trial, to estimate the causal relationship between a phenotype and an outcome variable. Our purpose is to extend the existing methods for such Mendelian randomization studies to the context of multiple genetic markers measured in multiple studies, based on the analysis of individual participant data. First, for a single genetic marker in one study, we show that the usual ratio of coefficients approach can be reformulated as a regression with heterogeneous error in the explanatory variable. This can be implemented using a Bayesian approach, which is next extended to include multiple genetic markers. We then propose a hierarchical model for undertaking a meta-analysis of multiple studies, in which it is not necessary that the same genetic markers are measured in each study. This provides an overall estimate of the causal relationship between the phenotype and the outcome, and an assessment of its heterogeneity across studies. As an example, we estimate the causal relationship of blood concentrations of C-reactive protein on fibrinogen levels using data from 11 studies. These methods provide a flexible framework for efficient estimation of causal relationships derived from multiple studies. Issues discussed include weak instrument bias, analysis of binary outcome data such as disease risk, missing genetic data, and the use of haplotypes.

Original language	English
Journal	Statistics in Medicine
Volume	29
Issue number	12
Pages (from-to)	1298-311
Number of pages	14
ISSN	0277-6715
DOIs	https://doi.org/10.1002/sim.3843
Publication status	Published - 30 May 2010

Access to Document

http://dx.doi.org/10.1002/sim.3843

Cite this

Burgess, S., Thompson, S. G., Thompson, G., Andrews, G., Samani, N. J., Hall-Andersen, A. M., Whincup, P., Morris, R., Lawlor, D. A., Davey Smith, G., Timpson, N., Ebrahim, S., Ben-Shlomo, Y., Brown, M., Ricketts, S., Sandhu, M., Reiner, A., Psaty, B., Lange, L., ... CRP CHD Genetics Collaboration (2010). Bayesian methods for meta-analysis of causal relationships estimated using genetic instrumental variables. Statistics in Medicine, 29(12), 1298-311. https://doi.org/10.1002/sim.3843

Burgess, S, Thompson, SG, Thompson, G, Andrews, G, Samani, NJ, Hall-Andersen, AM, Whincup, P, Morris, R, Lawlor, DA, Davey Smith, G, Timpson, N, Ebrahim, S, Ben-Shlomo, Y, Brown, M, Ricketts, S, Sandhu, M, Reiner, A, Psaty, B, Lange, L, Cushman, M, Hung, J, Thompson, P, Beilby, J, Warrington, N, Palmer, LJ, Nordestgaard, BG , Tybjærg-Hansen, A, Zacho, J, Wu, C, Lowe, G, Tzoulaki, I, Perera, MK, Yamamoto, JF, Chiodini, B, Franzosi, M, Hankey, GJ, Jamrozik, K, Palmer, L, Rimm, E, Pai, J, Heckbert, S, Bis, J, Anand, S, Engert, J, Collins, R, Clarke, R, Melander, O, Berglund, G, Ladenvall, P, Johansson, L & CRP CHD Genetics Collaboration 2010, 'Bayesian methods for meta-analysis of causal relationships estimated using genetic instrumental variables', Statistics in Medicine, vol. 29, no. 12, pp. 1298-311. https://doi.org/10.1002/sim.3843

@article{07dd2947ecab48c08f8cc8656c351e11,

title = "Bayesian methods for meta-analysis of causal relationships estimated using genetic instrumental variables",

abstract = "Genetic markers can be used as instrumental variables, in an analogous way to randomization in a clinical trial, to estimate the causal relationship between a phenotype and an outcome variable. Our purpose is to extend the existing methods for such Mendelian randomization studies to the context of multiple genetic markers measured in multiple studies, based on the analysis of individual participant data. First, for a single genetic marker in one study, we show that the usual ratio of coefficients approach can be reformulated as a regression with heterogeneous error in the explanatory variable. This can be implemented using a Bayesian approach, which is next extended to include multiple genetic markers. We then propose a hierarchical model for undertaking a meta-analysis of multiple studies, in which it is not necessary that the same genetic markers are measured in each study. This provides an overall estimate of the causal relationship between the phenotype and the outcome, and an assessment of its heterogeneity across studies. As an example, we estimate the causal relationship of blood concentrations of C-reactive protein on fibrinogen levels using data from 11 studies. These methods provide a flexible framework for efficient estimation of causal relationships derived from multiple studies. Issues discussed include weak instrument bias, analysis of binary outcome data such as disease risk, missing genetic data, and the use of haplotypes.",

author = "Stephen Burgess and Thompson, {Simon G} and Grahame Thompson and G Andrews and Samani, {N J} and Hall-Andersen, {Anne Mette} and P Whincup and R Morris and Lawlor, {D A} and {Davey Smith}, G and N Timpson and S Ebrahim and Y Ben-Shlomo and M Brown and S Ricketts and M Sandhu and A Reiner and B Psaty and L Lange and M Cushman and J Hung and P Thompson and J Beilby and N Warrington and Palmer, {L J} and Nordestgaard, {B G} and Anne Tybj{\ae}rg-Hansen and J Zacho and C Wu and G Lowe and I Tzoulaki and Perera, {Mahesha Kumari} and Yamamoto, {J F} and B Chiodini and M Franzosi and Hankey, {G J} and K Jamrozik and L Palmer and E Rimm and J Pai and S Heckbert and J Bis and S Anand and J Engert and R Collins and R Clarke and O Melander and G Berglund and P Ladenvall and L Johansson and Anne Tybj{\ae}rg-Hansen",

year = "2010",

month = may,

day = "30",

doi = "http://dx.doi.org/10.1002/sim.3843",

language = "English",

volume = "29",

pages = "1298--311",

journal = "Statistics in Medicine",

issn = "0277-6715",

publisher = "JohnWiley & Sons Ltd",

number = "12",

}

TY - JOUR

T1 - Bayesian methods for meta-analysis of causal relationships estimated using genetic instrumental variables

AU - Burgess, Stephen

AU - Thompson, Simon G

AU - Thompson, Grahame

AU - Andrews, G

AU - Samani, N J

AU - Hall-Andersen, Anne Mette

AU - Whincup, P

AU - Morris, R

AU - Lawlor, D A

AU - Davey Smith, G

AU - Timpson, N

AU - Ebrahim, S

AU - Ben-Shlomo, Y

AU - Brown, M

AU - Ricketts, S

AU - Sandhu, M

AU - Reiner, A

AU - Psaty, B

AU - Lange, L

AU - Cushman, M

AU - Hung, J

AU - Thompson, P

AU - Beilby, J

AU - Warrington, N

AU - Palmer, L J

AU - Nordestgaard, B G

AU - Tybjærg-Hansen, Anne

AU - Zacho, J

AU - Wu, C

AU - Lowe, G

AU - Tzoulaki, I

AU - Perera, Mahesha Kumari

AU - Yamamoto, J F

AU - Chiodini, B

AU - Franzosi, M

AU - Hankey, G J

AU - Jamrozik, K

AU - Palmer, L

AU - Rimm, E

AU - Pai, J

AU - Heckbert, S

AU - Bis, J

AU - Anand, S

AU - Engert, J

AU - Collins, R

AU - Clarke, R

AU - Melander, O

AU - Berglund, G

AU - Ladenvall, P

AU - Johansson, L

AU - CRP CHD Genetics Collaboration

PY - 2010/5/30

Y1 - 2010/5/30

N2 - Genetic markers can be used as instrumental variables, in an analogous way to randomization in a clinical trial, to estimate the causal relationship between a phenotype and an outcome variable. Our purpose is to extend the existing methods for such Mendelian randomization studies to the context of multiple genetic markers measured in multiple studies, based on the analysis of individual participant data. First, for a single genetic marker in one study, we show that the usual ratio of coefficients approach can be reformulated as a regression with heterogeneous error in the explanatory variable. This can be implemented using a Bayesian approach, which is next extended to include multiple genetic markers. We then propose a hierarchical model for undertaking a meta-analysis of multiple studies, in which it is not necessary that the same genetic markers are measured in each study. This provides an overall estimate of the causal relationship between the phenotype and the outcome, and an assessment of its heterogeneity across studies. As an example, we estimate the causal relationship of blood concentrations of C-reactive protein on fibrinogen levels using data from 11 studies. These methods provide a flexible framework for efficient estimation of causal relationships derived from multiple studies. Issues discussed include weak instrument bias, analysis of binary outcome data such as disease risk, missing genetic data, and the use of haplotypes.

AB - Genetic markers can be used as instrumental variables, in an analogous way to randomization in a clinical trial, to estimate the causal relationship between a phenotype and an outcome variable. Our purpose is to extend the existing methods for such Mendelian randomization studies to the context of multiple genetic markers measured in multiple studies, based on the analysis of individual participant data. First, for a single genetic marker in one study, we show that the usual ratio of coefficients approach can be reformulated as a regression with heterogeneous error in the explanatory variable. This can be implemented using a Bayesian approach, which is next extended to include multiple genetic markers. We then propose a hierarchical model for undertaking a meta-analysis of multiple studies, in which it is not necessary that the same genetic markers are measured in each study. This provides an overall estimate of the causal relationship between the phenotype and the outcome, and an assessment of its heterogeneity across studies. As an example, we estimate the causal relationship of blood concentrations of C-reactive protein on fibrinogen levels using data from 11 studies. These methods provide a flexible framework for efficient estimation of causal relationships derived from multiple studies. Issues discussed include weak instrument bias, analysis of binary outcome data such as disease risk, missing genetic data, and the use of haplotypes.

U2 - http://dx.doi.org/10.1002/sim.3843

DO - http://dx.doi.org/10.1002/sim.3843

M3 - Journal article

SN - 0277-6715

VL - 29

SP - 1298

EP - 1311

JO - Statistics in Medicine

JF - Statistics in Medicine

IS - 12

ER -

Bayesian methods for meta-analysis of causal relationships estimated using genetic instrumental variables

Abstract

Access to Document

Fingerprint

Cite this