Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “causal inference”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 91 records · Page 5Linked to original sources
PubMed2006

Instruments for causal inference: an epidemiologist's dream?

The use of instrumental variable (IV) methods is attractive because, even in the presence of unmeasured confounding, such methods may consistently estimate the average causal effect of an exposure on an outcome. However, for this consistent estimation to be achieved, several strong conditions must hold. We review the definition of an instrumental variable, describe the conditions required to obtain consistent estimates of causal effects, and explore their implications in the context of a recent application of the instrumental variables approach. We also present (1) a description of the connection between 4 causal models-counterfactuals, causal directed acyclic graphs, nonparametric structural equation models, and linear structural equation models-that have been used to describe instrumental variables methods; (2) a unified presentation of IV methods for the average causal effect in the study population through structural mean models; and (3) a discussion and new extensions of instrumental variables methods based on assumptions of monotonicity.

Bias↗
PubMed2004

An extended general location model for causal inferences from data subject to noncompliance and missing values.

Noncompliance is a common problem in experiments involving randomized assignment of treatments, and standard analyses based on intention-to-treat or treatment received have limitations. An attractive alternative is to estimate the Complier-Average Causal Effect (CACE), which is the average treatment effect for the subpopulation of subjects who would comply under either treatment (Angrist, Imbens, and Rubin, 1996, Journal of American Statistical Association 91, 444-472). We propose an extended general location model to estimate the CACE from data with noncompliance and missing data in the outcome and in baseline covariates. Models for both continuous and categorical outcomes and ignorable and latent ignorable (Frangakis and Rubin, 1999, Biometrika 86, 365-379) missing-data mechanisms are developed. Inferences for the models are based on the EM algorithm and Bayesian MCMC methods. We present results from simulations that investigate sensitivity to model assumptions and the influence of missing-data mechanism. We also apply the method to the data from a job search intervention for unemployed workers.

Algorithms↗
PubMed2000

Marginal structural models and causal inference in epidemiology.

In observational studies with exposures or treatments that vary over time, standard approaches for adjustment of confounding are biased when there exist time-dependent confounders that are also affected by previous treatment. This paper introduces marginal structural models, a new class of causal models that allow for improved adjustment of confounding in those situations. The parameters of a marginal structural model can be consistently estimated using a new class of estimators, the inverse-probability-of-treatment weighted estimators.

Anti-HIV Agents↗
PubMed1997

[Causal inference in medicine: a reaction to the report, "incidence of Minamata disease in communities along the Agano River, Niigata, Japan--patterns of the exposure and official diagnosis of patients"].

Kondo's "Incidence of Minamata Disease in Communities along the Agano River, Niigata, Japan (Jap. J. Hyg. 51:599-611;1996)" is critically reviewed. The data of the article were obtained from most of the residents living in the Agano river villages where Minamata disease was discovered in June, 1965. However, sampling proportions were much different between in the population base and in the cases. The method of identification of cases from the data and the reason for the difference were not clearly demonstrated. The citations of reference articles are insufficient despite the fact that other epidemiologic studies on methyl-mercury poisoning have been reported not only in Japan, but also around the world. His "analysis of the recognized patients" is erroneous. Both the sampling scheme of information of hair mercury and the modeling of the analysis are based on Kondo's arbitrary interpretation, not on epidemiologic theory. His "analysis of the rejected applicants" is also erroneous. His calculations of the attributable proportion are incorrect and self-induced in both the assignments of data and analysis of data. Kondo has failed to study the epidemiologic theories in light of changes in the field. Therefore, his article is lacking in epidemiologic theory, a logical base and scientific inference. In Japan, epidemiologic methodology has rarely been used in studies on Minamata Disease in either Kumamoto and Niigata. The government has used neurologically specific diagnosis based on combinations of symptoms to judge the causality between each of symptoms and methyl-mercury poisoning. Epidemiologic data obtained in Minamata, Kumamoto in 1971 indicate that the criteria set by the government in 1977 have produced much more false-negative patients than false-positive patients. As a result, a huge number of symptomatic patients, including those with peripheral neuropathy or with constriction of the visual field, did not receive any help or compensation until 1995. The authors emphasize that the causal relationship between each symptom and methyl-mercury exposure should be reevaluated epidemiologically in Japan.

Epidemiologic Methods↗
PubMed1995

Causal inference from indirect experiments.

An indirect experiment is a study in which randomized control is replaced by randomized encouragement, that is, subjects are encouraged, rather than forced, to receive a given treatment program. The purpose of this paper is to bring to the attention of experimental researchers simple mathematical results that enable us to assess, from indirect experiments, the strength with which causal influences operate among variables of interest. The results reveal that despite the laxity of the encouraging instrument, data from indirect experimentation can yield significant and sometimes accurate information on the impact of a program on the population as a whole, as well as on the particular individuals who participated in the program.

Humans↗
PubMed1999

Causal inference in a clinical trial: a comparative example.

Recently there has been much interest in methods for analyzing clinical trials of treatments that are subject to noncompliance. In this paper I study a small, simple dataset from a clinical trial of immunosuppressive therapy in the treatment of multiple sclerosis. I apply and compare a range of methods: the as-randomized (intention-to-treat) analysis, the as-treated analysis, estimates based on a nonignorable selection model, and Rubin's causal model. The results differ substantially even in this small dataset that exhibits modest noncompliance. For this reason, data analysts should be clear about which parameters are of greatest importance in the analysis of a clinical trial.

Azathioprine↗
PubMed2006

Causal inference of regulator-target pairs by gene mapping of expression phenotypes.

BACKGROUND: Correlations between polymorphic markers and observed phenotypes provide the basis for mapping traits in quantitative genetics. When the phenotype is gene expression, then loci involved in regulatory control can theoretically be implicated. Recent efforts to construct gene regulatory networks from genotype and gene expression data have shown that biologically relevant networks can be achieved from an integrative approach. In this paper, we consider the problem of identifying individual pairs of genes in a direct or indirect, causal, trans-acting relationship. RESULTS: Inspired by epistatic models of multi-locus quantitative trait (QTL) mapping, we propose a unified model of expression and genotype to identify quantitative trait genes (QTG) by extending the conventional linear model to include both genotype and expression of regulator genes and their interactions. The model provides mapping of specific genes in contrast to standard linkage approaches that implicate large QTL intervals typically containing tens of genes. In simulations, we found that the method can often detect weak trans-acting regulators amid the background noise of thousands of traits and is robust to transcription models containing multiple regulator genes. We reanalyze several pleiotropic loci derived from a large set of yeast matings and identify a likely alternative regulator not previously published. However, we also found that many regulators can not be so easily mapped due to the presence of cis-acting QTLs on the regulators, which induce close linkage among small neighborhoods of genes. QTG mapped regulator-target pairs linked to ARN1 were combined to form a regulatory module, which we observed to be highly enriched in iron homeostasis related genes and contained several causally directed links that had not been identified in other automatic reconstructions of that regulatory module. Finally, we also confirm the surprising, previously published results that regulators controlling gene expression are not enriched for transcription factors, but we do show that our more precise mapping model reveals functional enrichment for several other biological processes related to the regulation of the cell. CONCLUSION: By incorporating interacting expression and genotype, our QTG mapping method can identify specific regulator genes in contrast to standard QTL interval mapping. We have shown that the method can recover biologically significant regulator-target pairs and the approach leads to a general framework for inducing a regulatory module network topology of directed and undirected edges that can be used to identify leads in pathway analysis.

Algorithms↗
PubMed2026

Cell Type-Resolved Causal Inference and Spatial Transcriptomic Integration Reveal Immune-Specific Genetic Drivers of Autoimmune and Malignant Thyroid Disease.

BACKGROUND: Thyroid diseases, including autoimmune thyroid disease (AITD) and thyroid cancer, are characterized by immune dysregulation, yet the cell type-specific genetic mechanisms underlying these conditions remain poorly understood. Most genome-wide association studies (GWAS) have relied on bulk tissue expression quantitative trait loci (eQTL), which cannot resolve the heterogeneity of immune cell populations. METHODS: We performed two-sample Mendelian randomization (MR) analyses using single-cell cis-eQTLs from 14 immune cell subtypes (OneK1K cohort) as instrumental variables against GWAS summary statistics for four thyroid outcomes: autoimmune hyperthyroidism, autoimmune hypothyroidism, thyroid cancer and autoimmune thyroiditis. Causal associations were validated through Bayesian colocalization, phenome-wide association analysis (PheWAS) and multi-layered transcriptomic validation encompassing spatial transcriptomics of AITD tissue (GSE248205), bulk RNA-seq of thyroid cancer (GSE3678) and single-cell RNA-seq of thyroid tumours (GSE250521). gsMap spatial LD score regression was applied to map disease heritability onto spatial tissue architecture. RESULTS: We identified six Bonferroni-significant causal gene-cell type pairs for autoimmune hyperthyroidism, including protective effects of ABHD16A in na&#xef;ve/immature B cells (OR&#xa0;=&#xa0;0.440), HIST1H3H in CD8 NC T cells (OR&#xa0;=&#xa0;0.324), HMGN4 in NK recruiting cells (OR&#xa0;=&#xa0;0.556) and ZKSCAN4 in CD8 S100B T cells (OR&#xa0;=&#xa0;0.427), with five pairs showing strong colocalization (PP.H4 &#x2265; 86%). Three pairs reached significance for autoimmune hypothyroidism, including a risk association of HLA-F in CD4 NC T cells (OR&#xa0;=&#xa0;1.139). For autoimmune thyroiditis, FAM134B/RETREG1 showed consistent suggestive protective associations across both CD4 and CD8 NC T cells (PP.H4 &#x2265; 90% for both), suggesting a possible involvement of ER phagy regulation in thyroiditis susceptibility. Thyroid cancer showed a suggestive association with HLA-G in classical monocytes (OR&#xa0;=&#xa0;1.899, PP.H4&#xa0;=&#xa0;53%). Spatial transcriptomic validation demonstrated progressive immune infiltration from control tissue to Graves' disease to Hashimoto's thyroiditis (7.7%-15.7%, 46.1%-54.1%, respectively) and strong spatial correlation between target gene expression and corresponding cell type enrichment (e.g., plasma cell-HLA-DQB1: r&#xa0;=&#xa0;0.491, p < 10-300). HLA-G was independently validated in thyroid cancer bulk (log2fc&#xa0;=&#xa0;0.542, p&#xa0;=&#xa0;9.51&#xa0;&#xd7;&#xa0;10-3, AUC&#xa0;=&#xa0;0.857) and single-cell datasets. PheWAS revealed no significant associations detected for the core candidates. gsMap identified significant enrichment of autoimmune hypothyroidism heritability in gastrointestinal tract, adrenal gland and adipose tissue (all Bonferroni p < 0.002). CONCLUSIONS: This study establishes a multi-scale analytical framework integrating cell type-resolved genetic inference with spatial tissue validation, revealing distinct immunogenetic architectures underlying autoimmune versus malignant thyroid disease. Protective genetic programs in autoimmune hyperthyroidism converge on chromatin remodelling (HIST1H3H, HMGN4, ZKSCAN4) and lipid metabolism (ABHD16A) across lymphocyte subsets, whereas thyroid cancer risk involves immune escape mediated by HLA-G in myeloid cells. The ER-phagy receptor RETREG1 represents a candidate pathway warranting further investigation in autoimmune thyroiditis. These findings provide genetically supported, cell type-specific therapeutic targets and demonstrate a generalizable strategy for dissecting the immune-mediated mechanisms of complex thyroid diseases.

Mendelian randomization↗
PubMed2025

An Updated Polygenic Index Repository: Expanded Phenotypes, New Cohorts, and Improved Causal Inference.

Polygenic indexes (PGIs) - DNA-based predictors of individual phenotypes - have become essential tools across biomedical and social sciences. We introduce Version 2 of the Polygenic Index Repository, which expands phenotype coverage from 47 to 61, increases the number of participating datasets from 11 to 20, and adopts a more consistent and improved methodology for PGI construction. For 16 phenotypes, we leverage summary statistics from an updated GWAS meta-analysis with greater statistical power compared to the original release, thereby improving the PGI's predictive power. To improve power for family-based analyses, we provide imputed parental PGIs in all datasets with first-degree relatives and offer a framework for interpreting results from analyses that control for parental PGIs. We illustrate the utility of parental PGIs using two applications: (1) comparing PGI associations with and without parental PGI controls for all phenotypes in two Repository datasets with family data, and (2) for BMI and diastolic blood pressure, exploring the contribution of causal versus non-causal components of PGI associations to the imperfect portability of PGIs across subgroups within a genetic ancestry. Collectively, the updates enhance predictive performance, broaden the Repository's scope, and introduce novel resources that reduce confounding bias and improve interpretability.

Journal Article↗
PubMed2004

Sensitivity analysis of causal inference in a clinical trial subject to crossover.

In many clinical trials it is possible for some subjects to cross over between treatment arms. One can evaluate the effect of crossover by modeling it as a missing-data problem, where for subjects who cross over, one treats the unobserved value of the outcome in the original randomization arm as the missing data. The as-treated analysis is invalid if the crossover is nonignorable, in the sense that the crossovers represent a nonrandom sample of the randomized subjects. A recent area of general interest is the development of methods for measuring the sensitivity of inferences to nonignorability in the missing-data mechanism; one such approach is that of Troxel et al. In this paper we apply their method to the problem of measuring sensitivity to nonignorable crossover in randomized trials, extending it to the case where the crossover mechanism may differ between arms. Our method allows us to identify circumstances under which the as-treated analysis may be more or less sensitive to nonignorable crossover. We illustrate it with the example of a randomized clinical trial (RCT) in multiple sclerosis and a study of the effect of military service on income.

Causality↗
PubMed2026

Exploring China's Clean Air Act and associated cardiovascular disease risk: a prospective, quasi-experimental, and causal inference modelling study.

BACKGROUND: Substantial improvements in air quality have been recorded following the implementation of China's Clean Air Act (CCAA) in 2013. However, the association between CCAA implementation and individual-level cardiovascular disease (CVD) risk remains unclear. We aimed to examine the long-term association between CCAA implementation and individual-level predicted CVD risk. METHODS: In this prospective, quasi-experimental study, we used data from the China Kadoorie Biobank, a prospective cohort study that recruited participants from five urban and five rural areas across China between 2004 and 2008, with three resurveys conducted after the baseline survey (in 2008, 2013-14, and 2020-21). We included 34&#x2009;862 individuals (mean age 51&#xb7;3 years) who participated in at least one resurvey and had no history of CVD at baseline. Participants were classified into intervention (n=25&#x2009;497) and control (n=9365) groups based on the local government's targets for particulate matter reduction. We estimated the 10-year risk of incident CVD morbidity or mortality using a validated risk prediction model. We used a difference-in-difference model to assess the long-term association between CCAA implementation and predicted risk, with adjustments made for regional confounders and individual-level characteristics, including demographics, lifestyle factors, medical history, and indoor air pollution exposure. The relationship between changes in long-term exposure to PM2&#xb7;5, PM10, and O3 and predicted risk after CCAA implementation was analysed using a linear model. The estimated risk differences associated with air pollutant changes were estimated based on the magnitude of changes and their corresponding effect sizes. FINDINGS: After the CCAA was implemented, PM2&#xb7;5 and PM10 concentrations declined in both groups, but O3 concentrations increased. The intervention group showed a 3&#xb7;95% (95% CI 3&#xb7;18-4&#xb7;72%) lower increase in predicted risk than the control group, with larger estimated differences under stricter enforcement. Between 2013 and 2021, each 10 &#x3bc;g/m3 change in PM2&#xb7;5 concentration was positively associated with a 1&#xb7;80 (1&#xb7;34-2&#xb7;27) percentage point change in predicted CVD risk, whereas each 10 &#x3bc;g/m3 change in PM10 concentration was associated with a 1&#xb7;24 (0&#xb7;84-1&#xb7;63) percentage point change and each 10 &#x3bc;g/m3 change in O3 concentration with a 0&#xb7;58 (0&#xb7;33-0&#xb7;83) percentage point change. Overall, the observed changes in air pollutants during the study period were associated with an average 6&#xb7;6 percentage point reduction in predicted CVD risk. INTERPRETATION: The CCAA and improved air quality were associated with a slower increase in predicted CVD risk, supporting the necessity for stricter, multipollutant air quality policies to maximise public health benefits. FUNDING: National Natural Science Foundation of China, Kadoorie Charitable Foundation, Noncommunicable Chronic Diseases-National Science and Technology Major Project, National Key R&D Program of China, Chinese Ministry of Science and Technology, and UK Wellcome Trust.

Journal Article↗
PubMed1996

Inferring causality.

The goal of research is to discover new knowledge to improve patient care. Research is essential because of its ability to establish causal relationships. This article discusses the minimum requirements a research report must meet to establish a causal relationship.

Causality↗
PubMed2000

Predictive inference, causal reasoning, and model assessment in nonparametric Bayesian analysis: a case study.

This paper continues our earlier analysis of a data set on acute ear infections in small children, presented in Andreev and Arjas (1998). The main goal here is to provide a method, based on the use of predictive distributions, for assessing the possible causal influence which the type of day care will have on the incidence of ear infections. A closely related technique is used for the assessment of the nonparametric Bayesian intensity model applied in the paper. Two graphical methods, supported by formal tests, are suggested for this purpose.

Algorithms↗