ObjectiveTo analyze the causal relationship between the intake of cheese or tea and the risk of gastroesophageal reflux disease (GERD). MethodsUsing a two-sample Mendelian randomization approach, single nucleotide polymorphisms (SNPs) associated with milk or tea intake were used as instrumental variables. The causal effect of milk or tea intake on the risk of GERD was investigated using the MR Egger method, the weighted median method, the inverse-variance weighted (IVW) random-effects model, and the IVW fixed-effects model. Multivariable analysis was conducted using the MR Egger method, and leave-one-out sensitivity analysis was performed to validate the reliability of the data. ResultsCheese intake could reduce the occurrence of GERD [IVW random-effects model β=–1.010, 95%CI (0.265, 0.502), P<0.05], while tea intake could lead to the occurrence of GERD [IVW random-effects model β=0.288, 95%CI (1.062, 1.673), P<0.05]. ConclusionCheese intake may have a positive causal relationship with reducing the risk of GERD occurrence, while tea intake may have a positive causal relationship with increasing the risk of GERD occurrence.
Randomized controlled trial (RCT) are considered the "gold standard" for evaluating the causal effects of interventions on outcome measures. However, due to high research costs and ethical constraints, conducting RCT in clinical practice, especially in the surgical field, faces numerous challenges such as difficulties in subject recruitment, implementation of blinding, and standardization of interventions. In such cases, using real-world data to perform causal inference under the framework of target trial emulation (TTE), based on the principles of RCT design, helps to identify and reduce biases arising from design flaws in traditional observational studies, such as immortal time bias, confounding, selection bias, or collider bias. This approach can produce high-quality evidence comparable to that of RCT, thereby enhancing the clinical guidance value of real-world data studies. However, TTE has limitations, such as the inability to completely eliminate confounding, high quality requirements for source data, and the current lack of reporting standards. Therefore, researchers should be fully aware of these limitations to avoid making incorrect causal inferences. This article intends to provide an overview of the TTE framework, implementation points, application scope, application cases, and advantages and disadvantages of the framework.
In observational studies, multivariable analysis is commonly used to control confounding and reduce bias in the estimation of causal effect between exposure and outcome. However, in clinical problems with complex causal relationships, researchers select covariates for adjustment through clinical intuition and data-driven methods, which may lead to biased results. In recent years, directed acyclic graphs (DAGs) have become a popular method for visualizing causal relationships between variables. An appropriately constructed DAG can help researchers identify confounders, intermediate variables and other non-confounding variables, thereby improving covariates selection for multivariable analysis. In practice, researchers should incorporate clinical knowledge, systematic methods and transparent reporting to fully utilize DAG in causal inference, and support more reliable clinical decisions.
Objective To explore the causal association between radiation exposure and risk of head and neck cancer using Mendelian randomization (MR) method. Methods Genome-wide association studies of radiation exposure and head and neck cancer in the public database IEU OpenGWAS were identified, and single nucleotide polymorphisms (SNPs) were screened as instrumental variables. Two-sample MR analyses were performed using random-effect inverse variance weighted (IVW), fixed-effect IVW, weighted median, and MR-Egger methods to assess the causal association between radiation exposure and risk of head and neck cancer. Outliers were tested using the MR-PRESSO method, and heterogeneity was assessed using the Cochran Q test. MR-Egger regression intercept was utilized to detect gene-level pleiotropy, and a leave-one-out sensitivity analysis was conducted to evaluate the robustness of the study results. Results96 valid SNPs were included as instrumental variables. The analysis results of random-effect IVW method, fixed-effect IVW method, and weighted median method all showed that radiation exposure was associated with an increased risk of head and neck cancer [odds ratio and 95% confidence interval: 1.139 (1.065, 1.218), 1.139 (1.068, 1.215), and 1.141 (1.039, 1.253); P<0.05]. Heterogeneity testing did not reveal significant heterogeneity, MR-Egger regression analysis did not find gene level pleiotropy, and the leave-one-out method did not find a single SNP significantly affecting the overall estimation results. Conclusion Radiation exposure increases the risk of head and neck cancer, but this conclusion still needs further validation in more high-quality, large sample studies.