Causal Assumptions Research Articles

Under what conditions? A scenario‐based approach for exploring the prerequisites of future events

AbstractIn an increasingly complex world, futures thinking can be used to understand the conditions that define future events' realization. This paper presents a novel approach to explore under what conditions some chosen future event would occur. The approach can be seen as a fusion of exploratory scenarios with the backward‐looking perspective of backcasting while resembling cross‐impact methods in how future events' realization arises from the interaction of several uncertain assumptions. The approach maps the causal mechanisms and assumptions that lead to the investigated event's realization, and results in a subset of assumption combinations that lead to the event being realized or not, and scenario sets where the event is either realized or not. This analysis provides insights beyond “Is the event likely to occur?”, revealing the various circumstances under which it can occur. We also provide ways for considering probabilities and deliberate decisions within the approach. We illustrate the approach with three example cases from different problem domains, such as reaching long‐term climate targets, and employing various methods, such as causal mapping, influence diagrams, and optimization. We conclude with a discussion on the approach's potential to enhance foresight practice, emphasizing its synergistic relation to existing methods and its contribution to a richer, more nuanced anticipation of the future.

A critical evaluation of regression discontinuity studies in school effectiveness research

ABSTRACT School start regulations allocate children born immediately either side of a given date to different life paths: those slightly older starting school a full year earlier. School effectiveness literature exploits this to estimate causal effects described as ‘the absolute effect of schooling’ or ‘the effect of an additional year's schooling’, using the logic of regression discontinuity (RD). This paper examines the causal arguments and assumptions underpinning RD, noting particularly the importance of the causal description. It highlights concerns with describing causes in terms of school effectiveness including failure to consider the alternative treatment pathway; presence of other post-allocation causal factors and potential discontinuities at allocation. The paper notes that these can be overcome by using wider causal descriptions but at the expense of no longer identifying school effectiveness.

A causal inference framework for leveraging external controls in hybrid trials.

We consider the challenges associated with causal inference in settings where data from a randomized trial are augmented with control data from an external source to improve efficiency in estimating the average treatment effect (ATE). This question is motivated by the SUNFISH trial, which investigated the effect of risdiplam on motor function in patients with spinal muscular atrophy. While the original analysis used only data generated by the trial, we explore an alternative analysis incorporating external controls from the placebo arm of a historical trial. We cast the setting into a formal causal inference framework and show how these designs are characterized by a lack of full randomization to treatment and heightened dependency on modeling. To address this, we outline sufficient causal assumptions about the exchangeability between the internal and external controls to identify the ATE and establish a connection with novel graphical criteria. Furthermore, we propose estimators, review efficiency bounds, develop an approach for efficient doubly robust estimation even when unknown nuisance models are estimated with flexible machine learning methods, suggest model diagnostics, and demonstrate finite-sample performance of the methods through a simulation study. The ideas and methods are illustrated through their application to the SUNFISH trial, where we find that external controls can increase the efficiency of treatment effect estimation.

Making sense of feelings

Abstract Internal feeling states such as pain, hunger, and thirst are widely assumed to be drivers of behaviours essential for homeostasis and animal survival. Call this the ‘causal assumption’. It is becoming increasingly apparent that the causal assumption is incompatible with the standard view of motor action in neuroscience. While there is a well-known explanatory gap between neural activity and feelings, there is also a disjuncture in the reverse direction—what role, if any, do feelings play in animals if not to cause behaviour? To deny that feelings cause behaviours might thus seem to presage epiphenomenalism—the idea that subjective experiences, including feelings, are inert, emergent and, on some views, non-physical properties of brain processes. Since epiphenomenalism is antagonistic to fundamental commitments of evolutionary biology, the view developed here challenges the standard view about the function of feelings without denying that feelings have a function. Instead, we introduce the ‘sense making sense’ hypothesis—the idea that the function of subjective experience is not to cause behaviour, but to explain, in a restricted but still useful sense of ‘explanation’. A plausible framework is derived that integrates commonly accepted neural computations to blend motor control, feelings, and explanatory processes to make sense of the way feelings are integrated into our sense of how and why we do and what we do.

Fuzzy Regression Discontinuity Designs With Multiple Control Groups Under One-Sided Noncompliance: Evaluating Extended Time Accommodations

Since its conception by Thistlethwaite and Campbell, a regression discontinuity (RD) design has continuously evolved and progressed in various aspects. This paper explores a novel theoretical advancement in fuzzy RD designs by combining them with multiple control groups under one-sided noncompliance. Within this approach, we estimate the average treatment effect on the treated at the cutoff from the fuzzy RD design and construct its bounds derived from multiple control groups. Using the bounds as a sensitivity check, we examine whether the underlying causal or statistical assumptions for the fuzzy RD design are warranted. We verify the effectiveness of our approach via a simulation study and demonstrate its application by studying the effect of extended time accommodations using data from the National Assessment of Educational Progress.

Discussing the concept of substance-induced psychosis (SIP).

Substance-induced psychosis (SIP) is characterized by both substance use and a psychotic state, and it is assumed that the first causes the latter. In ICD-10 the diagnosis is categorized as and grouped together with substance use disorders, and to a large extent also treated as such in the health care system. Though criticism of the diagnostic construct of SIP dates back several decades, numerous large and high-quality studies have been published during the past 5-10 years that substantiate and amplify this critique. The way we understand SIP and even how we name it is of major importance for treatment and it has judicial consequences. It has been demonstrated that substance use alone is not sufficient to cause psychosis, and that other risk factors besides substance use are at play. These are risk factors that are also known to be associated with schizophrenia spectrum disorders. Furthermore, register-based studies from several different countries find that a large proportion, around one in four, of those who are initially diagnosed with an SIP over time are subsequently diagnosed with a schizophrenia spectrum disorder. This scoping review discusses the construct validity of SIP considering recent evidence. We challenge the immanent causal assumption in SIP, and advocate that the condition shares many features with the schizophrenia spectrum disorders. In conclusion, we argue that SIP just as well could be considered a first-episode psychotic disorder in patients with substance use.

A Systematic Approach to Evaluating Instrumental Variable Assumptions: Applied Example of Glucose-lowering Medications and Risk for Hospitalized Heart Failure in Older Adults.

One obstacle to adopting instrumental variable (IV) methods in pharmacoepidemiology is their reliance on strong, unverifiable assumptions. We can falsify IV assumptions by leveraging the causal structure, which can strengthen or refute their plausibility and increase the validity of effect estimates. We illustrate a systematic approach to evaluate calendar time IV assumptions in estimating the known effect of thiazolidinediones on hospitalized heart failure. Using cohort entry time before and after 09/2010, when the U.S. Food and Drug Administration issued a safety communication as a proposed IV, we estimated IV and propensity score-weighted 2-year risk differences (RDs) using Medicare data (2008-2014). We (i) performed inequality tests, (ii) identified the negative control IV/outcome using causal assumptions, (iii) estimated RDs after narrowing the calendar time range and excluding patients likely associated with unmeasured confounding, (iv) derived bounds for RDs, and (v) estimated the proportion of compliers and their characteristics. The findings revealed that IV assumptions were violated and RDs were extreme, but the assumptions became more plausible upon narrowing the calendar time range and restricting the cohort by excluding prevalent heart failure (the strongest measured predictor of outcome). Systematically evaluating IV assumptions could help detect bias in IV estimators and increase their validity.

102 - Rethinking the causal assumptions about maternal BMI, gestational weight gain and adverse pregnancy outcomes

102 - Rethinking the causal assumptions about maternal BMI, gestational weight gain and adverse pregnancy outcomes

Challenging unverified assumptions in causal claims: Do gas stoves increase risk of pediatric asthma?

The use of unverified models for risk estimates and policy recommendations can be highly misleading, as their predictions may not reflect real-world health impacts. For example, a recent article states that NO2 from gas stoves “likely causes ∼50,000 cases of current pediatric asthma from long-term NO2 exposure alone” annually in the United States. This explicitly causal claim, which is contrary to several methodology and review articles published in this journal, among others, reflects both (a) An unverified modeling assumption that pediatric asthma burden is approximately proportional to NO2; and (b) An unverified causal assumption that the assumed proportionality between exposure and response is causal. The article is devoid of any causal analysis showing that these assumptions are likely to be true. It does not show that reducing NO2 exposure from gas stoves would reduce pediatric asthma risk. Its key references report no significant associations – let alone causation – between NO2 and pediatric asthma. Thus, the underlying data suggests that the number of pediatric asthma cases caused by gas stoves in the United States is indistinguishable from zero. This highlights the need to rigorously validate modeling assumptions and causal claims in public health risk assessments to ensure scientifically sound foundations for policy decisions.

Methods for Extending Inferences From Observational Studies: Considering Causal Structures, Identification Assumptions, and Estimators.

Most prior work in quantitative approaches to generalizability and transportability emphasizes extending causal effect estimates from randomized trials to target populations. Extending findings from observational studies is also of scientific interest, and identifiability assumptions and estimation methods differ from randomized settings when there is selection on both the exposure and exposure-outcome mediators in combination with exposure-outcome confounders (and both confounders and mediators can modify exposure-outcome effects). We argue that this causal structure is common in observational studies, particularly in the field of life course epidemiology, for example, when extending estimates of the effect of an early-life exposure on a later-life outcome from a cohort enrolled in midlife or late life. We describe identifiability assumptions and identification using observed data in such settings, highlighting differences from work extending findings from randomized trials. We describe statistical methods, including weighting, outcome modeling, and doubly robust approaches, to estimate potential outcome means and average treatment effects in the target population and illustrate performance of the methods in a simulation study. We show that in the presence of selection into the study sample on both exposure and confounders, estimators must be able to address confounding in the target population. When there is also selection on mediators of the exposure-outcome relationship, estimators need to be able to use different sets of variables to account for selection (including the mediator), and confounding. We discuss conceptual implications of our results as well as highlight unresolved practical questions for applied work to extend findings from observational studies to target populations.

Path-specific causal decomposition analysis with multiple correlated mediator variables.

A causal decomposition analysis allows researchers to determine whether the difference in a health outcome between two groups can be attributed to a difference in each group's distribution of one or more modifiable mediator variables. With this knowledge, researchers and policymakers can focus on designing interventions that target these mediator variables. Existing methods for causal decomposition analysis either focus on one mediator variable or assume that each mediator variable is conditionally independent given the group label and the mediator-outcome confounders. In this article, we propose a flexible causal decomposition analysis method that can accommodate multiple correlated and interacting mediator variables, which are frequently seen in studies of health behaviors and studies of environmental pollutants. We extend a Monte Carlo-based causal decomposition analysis method to this setting by using a multivariate mediator model that can accommodate any combination of binary and continuous mediator variables. Furthermore, we state the causal assumptions needed to identify both joint and path-specific decomposition effects through each mediator variable. To illustrate the reduction in bias and confidence interval width of the decomposition effects under our proposed method, we perform a simulation study. We also apply our approach to examine whether differences in smoking status and dietary inflammation score explain any of the Black-White differences in incident diabetes using data from a national cohort study.

Process evaluation of a New psychosocial goal-setting and manualised support intervention for Independence in Dementia (NIDUS-Family).

We report a mixed-methods process evaluation embedded within a randomised controlled trial. We aimed to test and refine a theory of change model hypothesising key causal assumptions to understand how the New Interventions for Independence in Dementia Study (NIDUS)-Family (a manualised, multimodal psychosocial intervention), was effective relative to usual care, on the primary outcome of Goal Attainment Scaling (GAS) over 1year. In 2021-2022, intervention-arm dyads completed an acceptability questionnaire developed to test causal assumptions. We conducted qualitative interviews with dyads and intervention facilitators, purposively selected for diverse follow-up GAS scores. We collected observational data from intervention session recordings. We thematically analysed data, then integrated qualitative and quantitative data. 174/204 (85.3%) dyads allocated to NIDUS-Family, fully completed it, 18 partially completed, while 12 received no intervention. We interviewed 27/192 (14%) of dyads receiving any sessions, and 9/10 facilitators; and observed 12 sessions. 47/192 (24.5%) of carers completed the acceptability questionnaire. We identified four themes: (A) 'Someone to talk to helps dyads feel supported'; (B) 'NIDUS-Family helps carers change their perspective'; (C) 'Personalisation helps people living with dementia maintain their identity' and (D) 'Small steps help dyads move forward'. Key causal pathway mechanisms were: a respectful, trusting and impartial relationship with the facilitator: supporting the development of meaningful goals and support to find manageable solutions. Core implementation factors were delivery of the modules from a consistent facilitator across regular sessions. Core contextual factors influencing these mechanisms were dyadic participation and understanding of abilities.

How and why does it work? A video-based qualitative analysis of case conferences to reduce BPSD through the lens of Habermas’s theory of communicative action

BackgroundCase conferences are described as a goal-oriented, systematic method that team members can use to exchange professional opinions and develop treatment actions for a particular care problem. However, not all case conferences have proven to be effective. The Norwegian Targeted Interdisciplinary Model for the Evaluation and Treatment of Neuropsychiatric Symptoms (TIME) is an effective multicomponent model based on case conferences that informs approaches to behavioural and psychological symptoms in residents with dementia in nursing homes. Our aim was to explore how TIME case conferences structured based on cognitive behavioural therapy (CBT) contributed to person-centred actions and how the specific inductive structure of the TIME may have contributed to the effectiveness of the model.MethodsWe used video observation of six case conferences and analysed these videos by performing a thematic cross-case analysis of the transcripts from the videos and by iteratively watching the videos. According to Habermas’s theory of communicative action, we emphasized the case conference content, i.e., what was talked about in the case conferences, and the display of communication between the participants in the case conferences.ResultsOur findings showed that the theoretical principles behind the TIME, including both person-centred care and the inductive structure of CBT, reflected many aspects of Habermas’s theory of communicative actions. In particular, the TIME case conferences emphasized the lifeworld perspective for both residents and staff and contributed to what Habermas labelled communicative rationality as a means to develop shared understanding among staff and create person-centred action.ConclusionsOne causal assumption of how and why the TIME case conferences contributed to the effectiveness of the TIME in reducing BPSD in nursing homes is that the specific inductive structure of the case conferences with the column technique based on the ABC method together with PCC, emphasized the importance of the lifeworld for both the resident and the staff. Even though case conferences have been highlighted as useful, it is not indifferent how these case conferences are structured and conducted.Clinical trial registrationThe trial TIME was registered January 6, 2016, with clinicaltrials.gov (NCT02655003).

Overcoming Data Gaps in Life Course Epidemiology by Matching Across Cohorts.

Life course epidemiology is hampered by the absence of large studies with exposures and outcomes measured at different life stages in the same individuals. We describe when the effect of an exposure ( A ) on an outcome ( Y ) in a target population is identifiable in a combined ("synthetic") cohort created by pooling an early-life cohort including measures of A with a late-life cohort including measures of Y . We enumerate causal assumptions needed for unbiased effect estimation in the synthetic cohort and illustrate by simulating target populations under four causal models. From each target population, we randomly sampled early- and late-life cohorts and created a synthetic cohort by matching individuals from the two cohorts based on mediators and confounders. We estimated the effect of A on Y in the synthetic cohort, varying matching variables, the match ratio, and the strength of association between matching variables and A . Finally, we compared bias in the synthetic cohort estimates when matching variables did not d-separate A and Y to the bias expected in the original cohort. When the set of matching variables includes all variables d-connecting exposure and outcome (i.e., variables blocking all backdoor and front-door pathways), the synthetic cohort yields unbiased effect estimates. Even when matching variables did not fully account for confounders, the synthetic cohort estimate was sometimes less biased than comparable estimates in the original cohort. Methods based on merging cohorts may hasten the evaluation of early- and mid-life determinants of late-life health but rely on available measures of both confounders and mediators.

A Bayesian nonparametric approach for causal mediation with a post-treatment confounder.

We propose a new Bayesian nonparametric method for estimating the causal effects of mediation in the presence of a post-treatment confounder. The methodology is motivated by the Rural Lifestyle Intervention Treatment Effectiveness Trial (Rural LITE) for which there is interest in estimating causal mediation effects but is complicated by the presence of a post-treatment confounder. We specify an enriched Dirichlet process mixture (EDPM) to model the joint distribution of the observed data (outcome, mediator, post-treatment confounder, treatment, and baseline confounders). For identifiability, we use the extended version of the standard sequential ignorability (SI) as introduced in Hong etal. along with a Gaussian copula model assumption. The observed data model and causal identification assumptions enable us to estimate and identify the causal effects of mediation, that is, the natural direct effects (NDE) and natural indirect effects (NIE). Our method enables easy computation of NIE and NDE for a subset of confounding variables and addresses missing data through data augmentation under the assumption of ignorable missingness. We conduct simulation studies to assess the performance of our proposed method. Furthermore, we apply this approach to evaluate the causal mediation effect in the Rural LITE trial, finding that there was not strong evidence for the potential mediator.

Heat, humidity and health impacts: how causal diagrams can help tell the complex story

The global health burden associated with exposure to heat is a grave concern and is projected to further increase under climate change. While physiological studies have demonstrated the role of humidity alongside temperature in exacerbating heat stress for humans, epidemiological findings remain conflicted. Understanding the intricate relationships between heat, humidity, and health outcomes is crucial to inform adaptation and drive increased global climate change mitigation efforts. This article introduces ‘directed acyclic graphs’ (DAGs) as causal models to elucidate the analytical complexity in observational epidemiological studies that focus on humid-heat-related health impacts. DAGs are employed to delineate implicit assumptions often overlooked in such studies, depicting humidity as a confounder, mediator, or an effect modifier. We also discuss complexities arising from using composite indices, such as wet-bulb temperature. DAGs representing the health impacts associated with wet-bulb temperature help to understand the limitations in separating the individual effect of humidity from the perceived effect of wet-bulb temperature on health. General examples for regression models corresponding to each of the causal assumptions are also discussed. Our goal is not to prioritize one causal model but to discuss the causal models suitable for representing humid-heat health impacts and highlight the implications of selecting one model over another. We anticipate that the article will pave the way for future quantitative studies on the topic and motivate researchers to explicitly characterize the assumptions underlying their models with DAGs, facilitating accurate interpretations of the findings. This methodology is applicable to similarly complex compound events.

High-dimensional causal mediation analysis by partial sum statistic and sample splitting strategy in imaging genetics application.

Causal mediation analysis provides a systematic approach to explore the causal role of one or more mediators in the association between exposure and outcome. In omics or imaging data analysis, mediators are often high-dimensional, which brings new statistical challenges. Existing methods either violate causal assumptions or fail in interpretable variable selection. Additionally, mediators are often highly correlated, presenting difficulties in selecting and prioritizing top mediators. To address these issues, we develop a framework using Partial Sum Statistic and Sample Splitting Strategy, namely PS5, for high-dimensional causal mediation analysis. The method provides a powerful global mediation test satisfying causal assumptions, followed by an algorithm to select and prioritize active mediators with quantification of individual mediation contributions. We demonstrate its accurate type I error control, superior statistical power, reduced bias in mediation effect estimation, and accurate mediator selection using extensive simulations of varying levels of effect size, signal sparsity, and mediator correlations. Finally, we apply PS5 to an imaging genetics dataset of chronic obstructive pulmonary disease (COPD) patients ( N =8,897) in the COPDGene study to examine the causal mediation role of lung images ( p =5,810) in the associations between polygenic risk score and lung function and between smoking exposure and lung function, respectively. Both causal mediation analyses successfully estimate the global indirect effect and detect mediating image regions. Collectively, we find a region in the lower lobe of the right lung with a strong and concordant mediation effect for both genetic and environmental exposures. This suggests that targeted treatment toward this region might mitigate the severity of COPD due to genetic and smoking effects.

Revisiting the Transferability of Few-Shot Image Classification: A Frequency Spectrum Perspective.

Few-shot learning, especially few-shot image classification (FSIC), endeavors to recognize new categories using only a handful of labeled images by transferring knowledge from a model trained on base categories. Despite numerous efforts to address the challenge of deficient transferability caused by the distribution shift between the base and new classes, the fundamental principles remain a subject of debate. In this paper, we elucidate why a decline in performance occurs and what information is transferred during the testing phase, examining it from a frequency spectrum perspective. Specifically, we adopt causality on the frequency space for FSIC. With our causal assumption, non-causal frequencies (e.g., background knowledge) act as confounders between causal frequencies (e.g., object information) and predictions. Our experimental results reveal that different frequency components represent distinct semantics, and non-causal frequencies adversely affect transferability, resulting in suboptimal performance. Subsequently, we suggest a straightforward but potent approach, namely the Frequency Spectrum Mask (FRSM), to weight the frequency and mitigate the impact of non-causal frequencies. Extensive experiments demonstrate that the proposed FRSM method significantly enhanced the transferability of the FSIC model across nine testing datasets.

Multilevel Intervention Stepped Wedge Designs (MLI-SWDs)

Multilevel interventions (MLIs) hold promise for reducing health inequities by intervening at multiple types of social determinants of health consistent with the socioecological model of health. In spite of their potential, methodological challenges related to study design compounded by a lack of tools for sample size calculation inhibit their development. We help address this gap by proposing the Multilevel Intervention Stepped Wedge Design (MLI-SWD), a hybrid experimental design which combines cluster-level (CL) randomization using a Stepped Wedge design (SWD) with independent individual-level (IL) randomization. The MLI-SWD is suitable for MLIs where the IL intervention has a low risk of interference between individuals in the same cluster, and it enables estimation of the component IL and CL treatment effects, their interaction, and the combined intervention effect. The MLI-SWD accommodates cross-sectional and cohort designs as well as both incomplete (clusters are not observed in every study period) and complete observation patterns. We adapt recent work using generalized estimating equations for SWD sample size calculation to the multilevel setting and provide an R package for power and sample size calculation. Furthermore, motivated by our experiences with the ongoing NC Works 4 Health study, we consider how to apply the MLI-SWD when individuals join clusters over the course of the study. This situation arises when unemployment MLIs include IL interventions that are delivered while the individual is unemployed. This extension requires carefully considering whether the study interventions will satisfy additional causal assumptions but could permit randomization in new settings.

Rethinking causal assumptions about maternal BMI, gestational weight gain, and adverse pregnancy outcomes

BackgroundThe aim of this study was to evaluate commonly assumed causal relationships between body mass index (BMI), gestational weight gain (GWG), and adverse pregnancy outcomes, which have formed the basis of guidelines and interventions aimed at limiting GWG in women with overweight or obesity. We explored relationships between maternal BMI, total GWG (as a continuous variable and as ‘excessive’ GWG), and pregnancy outcomes (including infant birthweight measures and caesarean birth).MethodsAnalysis of individual participant data (IPD) from the i-WIP (International Weight Management in Pregnancy) Collaboration, from randomised trials of diet and/or physical activity interventions during pregnancy reporting GWG and maternal and neonatal outcomes.Women randomised to the control arm of 20 eligible randomised trials (4370 of 8908 participants) from the i-WIP dataset of 36 randomised trials (total 12,240 women). The main research questions were to characterise the relationship between maternal BMI and (a) total GWG, (b) the risk of ‘excessive’ GWG (using the Institute of Medicine’s guidelines), and (c) adverse pregnancy outcomes as mediated via GWG versus other pathways to determine the extent to which the observed effect of maternal BMI on pregnancy outcomes is mediated via GWG. We utilised generalised linear models and regression-based mediation analyses within an IPD meta-analysis framework.ResultsMean GWG decreased linearly as maternal BMI increased; however, the risk of ‘excessive’ GWG increased markedly at BMI category thresholds (i.e. between the normal and overweight BMI category threshold and between the overweight and obese BMI category threshold). Increasing maternal BMI was associated with increased risk of all pregnancy outcomes assessed; however, there was no evidence that this effect was mediated via effects on GWG.ConclusionsThere is evidence of a meaningful relationship between maternal BMI and GWG and between maternal BMI and adverse pregnancy outcomes. There is no evidence that the effect of maternal BMI on outcomes is via an effect on GWG. Our analyses also cast doubt on the existence of a relationship between ‘excessive’ GWG and adverse pregnancy outcomes. Our findings challenge the practice of actively managing GWG throughout pregnancy.