The number of Mendelian randomization (MR) analyses including large numbers of genetic variants is rapidly increasing. This is due to the proliferation of genome‐wide association studies, and the desire to obtain more precise estimates of causal effects. Since it is unlikely that all genetic variants will be valid instrumental variables, several robust methods have been proposed. We compare nine robust methods for MR based on summary data that can be implemented using standard statistical software. Methods were compared in three ways: by reviewing their theoretical properties, in an extensive simulation study, and in an empirical example. In the simulation study, the best method, judged by mean squared error was the contamination mixture method. This method had well‐controlled Type 1 error rates with up to 50% invalid instruments across a range of scenarios. Other methods performed well according to different metrics. Outlier‐robust methods had the narrowest confidence intervals in the empirical example. With isolated exceptions, all methods performed badly when over 50% of the variants were invalid instruments. Our recommendation for investigators is to perform a variety of robust methods that operate in different ways and rely on different assumptions for valid inferences to assess the reliability of MR analyses.

A large number of epidemiological studies use genetic variants as instrumental variables to infer causal relationships.^1,2 For a genetic variant to be a valid instrument in these so-called Mendelian randomization (MR) studies, three assumptions need to hold: (i) the genetic variant is associated with the exposure of interest (relevance assumption); (ii) the genetic variants should be independent of all confounders (independence assumption); (iii) the genetic variants only effect the outcome through the exposure of interest (exclusion restriction). Without specific knowledge about the biological mechanisms affected by genetic variants, it is virtually impossible to prove that the exclusion restriction holds for a specific genetic variant.³ For example, genetic variants may have pleiotropic effects on both the exposure and the outcome through different biological pathways.⁴

Background: Observational studies suggest that reproductive factors are associated with cardiovascular disease, but these are liable to influence by residual confounding. This study explores the causal relevance of reproductive factors on cardiovascular disease in women using Mendelian randomization. Methods and Results: Uncorrelated ( r ² <0.001), genome‐wide significant ( P <5×10 ⁻⁸ ) single‐nucleotide polymorphisms were extracted from sex‐specific genome‐wide association studies of age at first birth, number of live births, age at menarche, and age at menopause. Inverse‐variance weighted Mendelian randomization was used for primary analyses on outcomes of atrial fibrillation, coronary artery disease, heart failure, ischemic stroke, and stroke. Earlier genetically predicted age at first birth increased risk of coronary artery disease (odds ratio [OR] per year, 1.49 [95% CI, 1.28–1.74], P =3.72×10 ⁻⁷ ) heart failure (OR, 1.27 [95% CI, 1.06–1.53], P =0.009), and stroke (OR, 1.25 [95% CI, 1.00–1.56], P =0.048), with partial mediation through body mass index, type 2 diabetes, blood pressure, and cholesterol traits. Higher genetically predicted number of live births increased risk of atrial fibrillation (OR for 2 live births, 2.91 [95% CI, 1.16–7.29], P =0.023), heart failure (OR, 1.90 [95% CI, 1.28–2.82], P =0.001), ischemic stroke (OR, 1.86 [95% CI, 1.03–3.37], P =0.039), and stroke (OR, 2.07 [95% CI, 1.22–3.52], P =0.007). Earlier genetically predicted age at menarche increased risk of coronary artery disease (OR per year, 1.10 [95% CI, 1.06–1.14], P =1.68×10 ⁻⁶ ) and heart failure (OR, 1.12 [95% CI, 1.07–1.17], P =5.06×10 ⁻⁷ ); both associations were at least partly mediated by body mass index. Conclusions: These results support a causal role of a number of reproductive factors on cardiovascular disease in women and identify multiple modifiable mediators amenable to clinical intervention.

Background: Maternal cardiovascular risk factors have been associated with adverse maternal and fetal outcomes. Given the difficulty in establishing causal relationships using epidemiological data, we applied Mendelian randomization to explore the role of cardiovascular risk factors on risk of developing preeclampsia or eclampsia, and low fetal birthweight. Methods: Uncorrelated single-nucleotide polymorphisms associated systolic blood pressure (SBP), body mass index, type 2 diabetes, LDL (low-density lipoprotein) with cholesterol, smoking, urinary albumin-to-creatinine ratio, and estimated glomerular filtration rate at genome-wide significance in studies of 298 957 to 1 201 909 European ancestry participants were selected as instrumental variables. A 2-sample Mendelian randomization study was performed with primary outcome of preeclampsia or eclampsia (PET). Risk factors associated with PET were further investigated for their association with low birthweight. Results: Higher genetically predicted SBP was associated increased risk of PET (odds ratio [OR] per 1-SD SBP increase 1.90 [95% CI=1.45–2.49]; P =3.23×10 ⁻⁶ ) and reduced birthweight (OR=0.83 [95% CI=0.79–0.86]; P =3.96×10 ⁻¹⁸ ), and this was not mediated by PET. Body mass index and type 2 diabetes were also associated with PET (respectively, OR per 1-SD body mass index increase =1.67 [95% CI=1.44–1.94]; P =7.45×10 ⁻¹² ; and OR per logOR increase type 2 diabetes =1.11 [95% CI=1.04–1.19]; P =1.19×10 ⁻³ ), but not with reduced birthweight. Conclusions: Our results provide evidence for causal effects of SBP, body mass index, and type 2 diabetes on PET and identify that SBP is associated with reduced birthweight independently of PET. The results provide insight into the pathophysiological basis of PET and identify hypertension as a potentially modifiable risk factor amenable to therapeutic intervention.

It remains unclear whether the association between obstructive sleep apnoea (OSA), a form of sleep-disordered breathing (SDB), and atrial fibrillation (AF) is causal or mediated by shared co-morbidities such as obesity. Existing observational studies are conflicting and limited by confounding and reverse causality. We performed Mendelian randomisation (MR) to investigate the causal relationships between SDB, body mass index (BMI) and AF. Single-nucleotide polymorphisms associated with SDB (n = 29) and BMI (n = 453) were selected as instrumental variables to investigate the effects of SDB and BMI on AF, using genetic association data on 55,114 AF cases and 482,295 controls. Primary analysis was conducted using inverse-variance weighted MR. Higher genetically predicted SDB and BMI were associated with increased risk of AF (OR per log OR increase in snoring liability 2.09 (95% CI 1.10–3.98), p = 0.03; OR per 1-SD increase in BMI 1.33 (95% CI 1.24–1.42), p < 0.001). The association between SDB and AF was not observed in sensitivity analyses, whilst associations between BMI and AF remained consistent. Similarly, in multivariable MR, SDB was not associated with AF after adjusting for BMI (OR 0.68 (95% CI 0.42–1.10), p = 0.12). Higher BMI remained associated with increased risk of AF after adjusting for OSA (OR 1.40 (95% CI 1.30–1.51), p < 0.001). Elevated BMI appears causal for AF, independent of SDB. Our data suggest that the association between SDB, in general, and AF is attributable to mediation or confounding from obesity, though we cannot exclude that more severe SDB phenotypes (i.e., OSA) are causal for AF.