SEARCH · Search PubMed
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.
Some purely methodological comments are made on the pitfalls and difficulties in making causal inferences from observational data, including in studies of disparity in medicine. The ideas of spurious correlation and measurement error are discussed with an eye towards their impact upon inferences about causality, and cautions are offered about over-reliance upon testing hypotheses.
Research suggests that by the age of five, children have extensive causal knowledge, in the form of intuitive theories. The crucial question for developmental cognitive science is how young children are able to learn causal structure from evidence. Recently, researchers in computer science and statistics have developed representations (causal Bayes nets) and learning algorithms to infer causal structure from evidence. Here we explore evidence suggesting that infants and children have the prerequisites for making causal inferences consistent with causal Bayes net learning algorithms. Specifically, we look at infants and children's ability to learn from evidence in the form of conditional probabilities, interventions and combinations of the two.
Specificity of association between putative risk factor and disease under study is important to inference on causality. Nevertheless many studies investigate mortality of a single disease without comparison with a control. Age-standardized proportional mortality ratios make single disease studies into case-control studies and thus demonstrate whether or not associations are disease specific. Comparison of disease-specific with all-cause mortality experiences of whole populations classified by exposure, clearly distinguishes between exposures associated with more death and with earlier/younger death, thereby overcoming an important limitation of the familiar standardized mortality ratio (SMR). Smoking is associated with more death from lung cancer (lifetime cause-specific proportions, never 1%, light 6%, moderate 8% and heavy 12%) and with earlier/younger death from ischaemic heart disease (never 35%, light 34%, moderate 32% and heavy 29%).
Previous research has suggested that preschoolers possess a cognitive system that allows them to construct an abstract, coherent representation of causal relations among events. Such a system lets children reason retrospectively when they observe ambiguous data in a rational manner (e.g., D. M. Sobel, J. B. Tenenbaum, & A. Gopnik, 2004). However, there is little evidence that demonstrates whether younger children possess similar inferential abilities. In Experiment 1, the authors extended previous findings with older children to examine 19- and 24-month-olds' causal inferences. Twenty-four-month-olds' inferences were similar to those of preschoolers, but younger children lacked the ability to make retrospective causal inferences, perhaps because of performance limitations. In Experiment 2, the authors designed an eye-tracking paradigm to test younger participants that eliminated various manual search demands. Eight-month-olds' anticipatory eye movements, in response to retrospective data, revealed inferences similar to those of 24-month-olds in Experiment 1 and preschoolers in previous research. These data are discussed in terms of associative reasoning and causal inference.
BACKGROUND: Hormone replacement therapy (HRT) is widely prescribed for the management of hormone deficiency, particularly during menopause, yet its causal effects on human brain structure remain incompletely understood. Observational studies have reported heterogeneous associations, underscoring the need for robust causal inference. METHODS: We applied an integrated causal framework combining two-sample Mendelian Randomization (MR) and Double Machine Learning (DML) to evaluate the effects of four HRT-related exposures-age at initiation, age at cessation, ever-use of HRT, and a composite medication-based phenotype-on 1366 brain imaging-derived phenotypes from the UK Biobank. Genetic instruments were derived from large-scale GWAS summary statistics, and causal estimates were validated using non-parametric DML models with cross-fitting and performance evaluation. RESULTS: Genetic instruments for age at HRT initiation, age at cessation, and ever-use of HRT were strong (median F-statistics 16.29-36.66). MR analyses identified a causal association between later initiation of HRT and lower orientation dispersion in the right inferior cerebellar peduncle (ubm-a-542; primary finding, no pleiotropy detected). An additional association with the left tapetum FA (ubm-a-243) was identified but exhibited significant directional horizontal pleiotropy (MR-Egger intercept P = 0.001) and is excluded from primary conclusions (Supplementary Note S2). Later cessation of HRT was associated with increased cortical thickness in the left middle occipital gyrus, reduced surface area in the left frontopolar cortex, and increased orientation dispersion in the splenium of the corpus callosum. Ever-use of HRT was causally linked to larger volumes of the right inferior frontal gyrus and right nucleus accumbens. These associations were corroborated by independent DML validation, which provided causally debiased estimates robust to high-dimensional confounding. Results for ukb-b-8080 (median F = 1.45) are provided in Supplementary Note S1 only; weak-instrument bias precludes causal inference. CONCLUSIONS: This study provides genetic-instrument-based and machine-learning-validated evidence for causal associations between HRT exposure-particularly its timing and lifetime use-and specific features of human brain structure, including white-matter microarchitecture, cortical thickness, and regional brain volume. These findings are FDR-controlled within exposures and independently replicated by DML, but require replication in external neuroimaging GWAS cohorts to establish definitive causal conclusions. They highlight the neurobiological relevance of sex steroid exposure and inform future research on brain aging and personalized hormone-based interventions.
Five studies investigated (a) children's ability to use the dependent and independent probabilities of events to make causal inferences and (b) the interaction between such inferences and domain-specific knowledge. In Experiment 1, preschoolers used patterns of dependence and independence to make accurate causal inferences in the domains of biology and psychology. Experiment 2 replicated the results in the domain of biology with a more complex pattern of conditional dependencies. In Experiment 3, children used evidence about patterns of dependence and independence to craft novel interventions across domains. In Experiments 4 and 5, children's sensitivity to patterns of dependence was pitted against their domain-specific knowledge. Children used conditional probabilities to make accurate causal inferences even when asked to violate domain boundaries.
BACKGROUND: Two recent accounts of the use of causal criteria make opposite claims: that criteria should be used more often to avoid bias in assessments of weak associations and, in direct contrast, that criteria are scientifically invalid. METHODS: A recent review of the current practice of causal inference in epidemiology, as well as some more theoretical concerns, reveals errors in the two claims. RESULTS: In practice, epidemiologists often use the criteria of consistency, strength, dose-response, and biological plausibility, but not often temporality, when judging weak associations. These criteria are used for causal assessments as well as for making public health recommendations. In theory, causal criteria can be used to either refute or predict causal effects. CONCLUSION: Research on causal inference methodology should be encouraged, including research on underlying theory, methodology, and additional systematic descriptions of how causal inference is practised. Specific research questions include: to what extent can consensus be achieved on definitions and accompanying rules of inference for criteria, the relationship of meta-analysis to the criterion of consistency, and the interrelationships of criteria such as consistency, strength of association, and biological plausibility.
This paper proposes a theoretical synthesis of the concepts of organizational size, technology, complexity, and structural differentiation. It suggests and finds that the most important determinant of differentiation in the division of labor is the scope of an organization's task, a technological dimension, and not organizational size. Neither horizontal nor vertical differentiation is thought to be determined by size while the scope of the task is proposed as a determinant of horizontal differentiation. The findings, however, support only the inference of a moderate causal connection between either size or task scope and either form of differentiation. Throughout the analysis and discussion the contrast between causal inference based on associations of levels and change rates is discussed. When dealing with the organizational processes addressed in this paper, both associations must be examined to adequately portray the complexity of the causal processes inferred.
Advancing understanding of the importance of breastfeeding (BF) has required advances in application of epidemiologic methods to BF research. This chapter considers, in the context of BF epidemiology, how different research designs complement each other to establish firmer causal inferences, the need for multidisciplinary collaborations, and the application of powerful analytical methods to strengthen epidemiologic studies. A recent global meta-analysis of 47 studies documenting the inverse association between lifetime BF duration and breast cancer risk illustrates the value of longitudinal (cohort) studies to establish the temporal sequence of events and to rule out differential recall biases that can threaten the internal validity of case-control studies. The discovery of the transmission of human immunodeficiency virus (HIV) via human milk has led epidemiologists to provide important hypotheses to be tested at the molecular and cellular level and in turn test hypotheses generated in laboratory studies. Application of powerful statistical methods to BF epidemiology has helped strengthen causal inferences. A retrospective analysis of the 1976--1977 Malaysian Family Life Survey found interactive multivariate statistical models to better understand how different socioeconomic and cultural contexts modify the influence of BF on child health. Further, a study conducted in Cebu City, Philippines, demonstrated how structural equation models can be used to test the pathways through which proximate (e.g., BF) and distal (e.g., socioeconomic status) determinants affect child health after adjusting for reverse causality (i.e., child health influencing maternal BF choice). For ethical reasons, BF randomized controlled trials (RCTs) have seldom been conducted. Recent RCTs examining the link between BF peer counseling and BF outcomes in diverse settings provides a unique opportunity to firmly establish the causal inferences regarding BF and the prevention of diarrheal diseases. The value of rapid assessment anthropologic methods to guide development of peer counseling interventions subsequently tested through RCTs is presented. This chapter also illustrates how Cox regression and logistic regression can be applied to BF research.
In the shadowy world between philosophy of science and ethics lie the paired concepts of underdetermination and incommensurability. Typically, scientific evidence underdetermines the hypotheses tested in research studies, providing neither proof nor disproof. As a result, scientists must judge the weight of the evidence, and in doing so, bring scientific and extrascientific values to bear in their approaches to assessing and interpreting the evidence. When different scientists employ very different values, their views are said to be incommensurable. Less prominent differences represent partial incommensurabilities. The definitions and analyses provided by McMullin and by Veatch and Stempsey lay the foundation for the description of partial incommensurabilities in the current practice of assessing and interpreting epidemiologic evidence. This practice is called "causal inference" and is undertaken for the purpose of making causal conclusions and public health recommendations from population-based studies of exposures and diseases. Following the work of Bayley and Longino, several suggestions are examined for dealing with the partial incommensurabilities found in the general practice of causal inference in contemporary epidemiology. Two specific examples illustrate these ideas: studies on the relationship between induced abortion and breast cancer and those on the relationship between moderate alcohol consumption and breast cancer.
Epidemiological methods, which combine population thinking and group comparisons, can primarily identify causes of disease in populations. There is therefore a tension between our intuitive notion of a cause, which we want to be deterministic and invariant at the individual level, and the epidemiological notion of causes, which are invariant only at the population level. Epidemiologists have given heretofore a pragmatic solution to this tension. Causal inference in epidemiology consists in checking the logical coherence of a causality statement and determining whether what has been found grossly contradicts what we think we already know: how strong is the association? Is there a dose-response relationship? Does the cause precede the effect? Is the effect biologically plausible? Etc. This approach to causal inference can be traced back to the English philosophers David Hume and John Stuart Mill. On the other hand, the mode of establishing causality, devised by Jakob Henle and Robert Koch, which has been fruitful in bacteriology, requires that in every instance the effect invariably follows the cause (e.g., inoculation of Koch bacillus and tuberculosis). This is incompatible with epidemiological causality which has to deal with probabilistic effects (e.g., smoking and lung cancer), and is therefore invariant only for the population.
The role of nutritional epidemiology studies in the development of nutritional recommendations has been controversial, in part because individual studies supporting either side of a given issue can often be identified. Several sets of criteria for inference of a causal relation between a dietary factor and a disease from epidemiologic studies have been suggested. One such set is that of Sir Austin Bradford Hill, which includes criteria such as strength of association, dose-response relation, consistency of association, temporally correct association, specificity of association, and biological plausibility. Another set of criteria, used by the US Preventive Services Task Force, ranks evidence according to study design, designating evidence from randomized controlled trials as superior to evidence from cohort or case-control studies, which are in turn superior to evidence from ecologic studies or opinions of respected authorities. The application of these criteria to the question of whether vitamin E intake is associated with coronary heart disease is examined here. It is suggested that the epidemiologic evidence from prospective cohort studies generally supports an inverse association of vitamin E intake and risk of coronary heart disease. The information available from randomized trials is limited but suggestive of an inverse association with nonfatal, but not with fatal, coronary events. It is suggested that the application of criteria for causal inference to specific questions in nutritional epidemiology may provide clarity to seemingly contradictory information.
In this paper, criteria used by many epidemiologists as aids in causal inference are reviewed and revised. The revised scheme emphasizes the distinction between essential properties of a cause and criteria useful for deciding on the presence of these properties in a given case. A systematic procedure for causal inference tests each essential causal property in turn against appropriate criteria. For a pragmatic epidemiology in which all determinants serve as causes, their essential properties are held to be association, time order, and direction, in an ascending hierarchy. Criteria for association are probabilistic and can be enhanced by strength and consistency. Given association, criteria for time order of the relevant variables follow from access to observation, which is dependent on design. Given association and time order, causal direction (or consequential change) calls on an array of criteria, namely, consistency and survivability, strength, specificity in cause and in effect, predictive performance, and coherence in all its forms (e.g., theoretical, factual, biologic, and statistical). The evolution of such criteria is traced through the epidemiologic literature in the light of historical context. Although Popper's philosophy cannot directly serve an inherently inductive judgmental process, his notion of survivability has here been added, alongside replicability, as a subclass of consistency. This criterion is proposed to bridge the gap between the particularity of designs and the generality required of causal relations. Designs are ordered and described in the framework of testing survivability. Finally, definitions are offered for the list of criteria deployed.
The medical community is forced to accelerate the move from opinion-based to evidence-based medicine, that is, to aim at basing all caring and clinical practice on empiri. A clear-cut epistemology would facilitate this process. In this article we present a comprehensive clinical epidemiological theory which can be used for validity issues in caring science, quality of life research, controlled clinical trials and compilations of uncontrolled evidence. The theory is based on four distinct stages that can be identified in a study, whereof the first is demarcation of the source person-time. A source person-time ('study base') can be identified for any study in all disciplines, giving an argument for using this concept as the common reference point for validity issues. Apart from identifying the source person-time, recovery of the actually observed person-time, collection of data and calculation of an ('adjusted') effect parameter (e.g., incidence ratio) are additional stages of a study. When the source person-time is demarcated confounding is introduced, when the actually observed person-time is recovered misrepresentation, in the third stage misclassification and in the fourth analytical alteration of the parameter of effect. The concept of the source person-time can, in addition, link examination of validity in caring and clinical sciences to observational studies, thereby allowing the field to benefit from all theoretical achievements for preventing, handling and understanding systematic errors developed in epidemiology. We conclude it is possible to implement a common terminology of validity for all caring and medical sciences. Drawing causal inferences in these disciplines is not mechanical, it can never, for example, be done with statistical inference. Establishing a causal relation always includes an assessment of the magnitude and direction of systematic errors influencing the adjusted effect parameter. From the presented epistemology it follows that differences in validity from a case history to a large randomized, placebo-controlled and double-blinded study are quantitative rather than qualitative. This realization in turn opens up for a more refined discussion of when a decision is evidence-based rather than opinion-based.
Causal inference of exposure-response relations from data is a challenging aspect of risk assessment with important implications for public and private risk management. Such inference, which is fundamentally empirical and based on exposure (or dose)-response models, seldom arises from a single set of data; rather, it requires integrating heterogeneous information from diverse sources and disciplines including epidemiology, toxicology, and cell and molecular biology. The causal aspects we discuss focus on these three aspects: drawing sound inferences about causal relations from one or more observational studies; addressing and resolving biases that can affect a single multivariate empirical exposure-response study; and applying the results from these considerations to the microbiological risk management of human health risks and benefits of a ban on antibiotic use in animals, in the context of banning enrofloxacin or macrolides, antibiotics used against bacterial illnesses in poultry, and the effects of such bans on changing the risk of human food-borne campylobacteriosis infections. The purposes of this paper are to describe novel causal methods for assessing empirical causation and inference; exemplify how to deal with biases that routinely arise in multivariate exposure- or dose-response modeling; and provide a simplified discussion of a case study of causal inference using microbial risk analysis as an example. The case study supports the conclusion that the human health benefits from a ban are unlikely to be greater than the excess human health risks that it could create, even when accounting for uncertainty. We conclude that quantitative causal analysis of risks is a preferable to qualitative assessments because it does not involve unjustified loss of information and is sound under the inferential use of risk results by management.
The last decade saw enormous progress in the development of causal inference tools to account for noncompliance in randomized clinical trials. With survival outcomes, structural accelerated failure time (SAFT) models enable causal estimation of effects of observed treatments without making direct assumptions on the compliance selection mechanism. The traditional proportional hazards model has however rarely been used for causal inference. The estimator proposed by Loeys and Goetghebeur (2003, Biometrics vol. 59 pp. 100-105) is limited to the setting of all or nothing exposure. In this paper, we propose an estimation procedure for more general causal proportional hazards models linking the distribution of potential treatment-free survival times to the distribution of observed survival times via observed (time-constant) exposures. Specifically, we first build models for observed exposure-specific survival times. Next, using the proposed causal proportional hazards model, the exposure-specific survival distributions are backtransformed to their treatment-free counterparts, to obtain - after proper mixing - the unconditional treatment-free survival distribution. Estimation of the parameter(s) in the causal model is then based on minimizing a test statistic for equality in backtransformed survival distributions between randomized arms.
Social factors are associated with a wide variety of health outcomes. Social epidemiology has successfully used the traditional methods of surveillance and description to establish consistent relations between social factors and health status. Epidemiology as an etiologic science, however, has been largely ineffective in moving toward causal explanations for these observed patterns. Using the counterfactual approach to causal inference, the authors describe several fundamental problems that often arise when researchers seek to infer explanatory mechanisms from data on social factors. Contrasts that form standard causal effect estimates require implicit unobserved (counterfactual) quantities, because observational data provide only one exposure state for each individual. Although application of counterfactual arguments has successfully advanced etiologic understanding in other observational settings, the particular nature of social factors often leads to logical contradictions or misleading inferences when investigators fail to clearly articulate the counterfactual contrasts that are implied. For example, because social factors are often attributes of individuals and are components of structured social relations, random assignment is not plausible even as a hypothetical experiment, making typical epidemiologic contrasts inappropriate and the inference equivocal at best. Accordingly, more deliberate and creative approaches to causal inference in social epidemiology are required. Infectious disease epidemiology and systems analysis provide examples of approaches to causal inference that can be used when statistical mimicry of simple experimental designs is not tenable. In an era of increasing social inequality, valid approaches for the study of social factors and health are needed more urgently than ever.
Eight experiments evaluated a core assumption of several theories of text processing, the shared resource assumption, which states that component text processes share limited processing resources. Short texts each contained two critical sentences that together warranted a causal inference. The syntactic structure of the second sentence was either more or less difficult to parse. Results from a lexical decision task suggested that readers formed the causal inferences when the syntactic structure was less difficult to parse but that inferencing was constrained when syntactic structure was more difficult. Follow-up experiments suggested that this interference was not due to inferior output of the syntactic parser nor to the increased demands of difficult syntax interfering with maintenance of information needed to form the inference. The results suggest insufficient resources were available for the operation of inference processes due to the increased demands of syntactic parsing, consistent with the shared resource assumption.