Potential Outcomes Framework Research Articles

Sugar-sweetened beverage intake and convenience store shopping as mediators of the food insecurity-Tooth decay relationship among low-income children in Washington state.

There are oral health disparities in the U.S. and children in food-insecure households have a higher burden of tooth decay. Identifying the mechanisms underlying the food insecurity-tooth decay relationship could inform public health interventions. This study examined how sugar-sweetened beverage (SSB) intake and frequent convenience store shopping mediated the food insecurity-tooth decay relationship for lower-income children. Cross-sectional study data included a household survey, beverage questionnaire, and dental examination. The sample included 452 lower-income, racially-diverse, child-caregiver dyads in 2018 from King County in Washington state. The exposure was household food insecurity, the outcome was untreated decayed tooth surfaces, and the proposed mediators were SSB intake and frequent convenience store shopping (≥2 times/week). Causal mediation analyses via the potential outcomes framework was used to estimate natural indirect and direct effects. Fifty-five percent of participants were in food-insecure households, the mean number of decayed tooth surfaces among children was 0.87 (standard deviation [SD] = 1.99), the mean SSB intake was 17 fluid ounces (fl/oz)/day (SD = 35), and 18% of households frequently shopped at a convenience store. After adjusting for confounders, household food insecurity and log-transformed SSB intake (fluid ounces/day) were positively associated with decayed tooth surfaces, but not at the a α = 0.05 level (mean ratio [MR] 1.60; 95% confidence interval [CI] 0.89, 2.88; p = .12 and MR 1.16; 95% CI 0.93, 1.46; p = .19, respectively). Frequent convenience store shopping was associated with 2.75 times more decayed tooth surfaces (95% CI 1.61, 4.67; p < .001). SSB intake mediated 10% of the food insecurity-tooth decay relationship (p = .35) and frequent convenience store shopping mediated 22% (p = .33). Interventions aimed at addressing oral health disparities in children in food-insecure households could potentially focus on reducing intake of SSBs and improving access to healthful foods in lower-income communities.

Evaluation of randomized controlled trials: a primer and tutorial for mental health researchers

BackgroundConsidered one of the highest levels of evidence, results of randomized controlled trials (RCTs) remain an essential building block in mental health research. They are frequently used to confirm that an intervention “works” and to guide treatment decisions. Given their importance in the field, it is concerning that the quality of many RCT evaluations in mental health research remains poor. Common errors range from inadequate missing data handling and inappropriate analyses (e.g., baseline randomization tests or analyses of within-group changes) to unduly interpretations of trial results and insufficient reporting. These deficiencies pose a threat to the robustness of mental health research and its impact on patient care. Many of these issues may be avoided in the future if mental health researchers are provided with a better understanding of what constitutes a high-quality RCT evaluation.MethodsIn this primer article, we give an introduction to core concepts and caveats of clinical trial evaluations in mental health research. We also show how to implement current best practices using open-source statistical software.ResultsDrawing on Rubin’s potential outcome framework, we describe that RCTs put us in a privileged position to study causality by ensuring that the potential outcomes of the randomized groups become exchangeable. We discuss how missing data can threaten the validity of our results if dropouts systematically differ from non-dropouts, introduce trial estimands as a way to co-align analyses with the goals of the evaluation, and explain how to set up an appropriate analysis model to test the treatment effect at one or several assessment points. A novice-friendly tutorial is provided alongside this primer. It lays out concepts in greater detail and showcases how to implement techniques using the statistical software R, based on a real-world RCT dataset.DiscussionMany problems of RCTs already arise at the design stage, and we examine some avoidable and unavoidable “weak spots” of this design in mental health research. For instance, we discuss how lack of prospective registration can give way to issues like outcome switching and selective reporting, how allegiance biases can inflate effect estimates, review recommendations and challenges in blinding patients in mental health RCTs, and describe problems arising from underpowered trials. Lastly, we discuss why not all randomized trials necessarily have a limited external validity and examine how RCTs relate to ongoing efforts to personalize mental health care.

Bayesian causal inference for observational studies with missingness in covariates and outcomes.

Missing data are a pervasive issue in observational studies using electronic health records or patient registries. It presents unique challenges for statistical inference, especially causal inference. Inappropriately handling missing data in causal inference could potentially bias causal estimation. Besides missing data problems, observational health data structures typically have mixed-type variables - continuous and categorical covariates - whose joint distribution is often too complex to be modeled by simple parametric models. The existence of missing values in covariates and outcomes makes the causal inference even more challenging, while most standard causal inference approaches assume fully observed data or start their works after imputing missing values in a separate preprocessing stage. To address these problems, we introduce a Bayesian nonparametric causal model to estimate causal effects with missing data. The proposed approach can simultaneously impute missing values, account for multiple outcomes, and estimate causal effects under the potential outcomes framework. We provide three simulation studies to show the performance of our proposed method under complicated data settings whose features are similar to our case studies. For example, Simulation Study 3 assumes the case where missing values exist in both outcomes and covariates. Two case studies were conducted applying our method to evaluate the comparative effectiveness of treatments for chronic disease management in juvenile idiopathic arthritis and cysticfibrosis.

Bayesian Inference of Dynamic Mediation Models for Longitudinal Data

Mediation analysis is widely applied in various fields of science, such as psychology, epidemiology, and sociology. In practice, many psychological and behavioral phenomena are dynamic, and the corresponding mediation effects are expected to change over time. However, most existing mediation methods assume a static mediation effect over time, which overlooks the dynamic nature of mediation effect. To address this issue, we propose dynamic mediation models that can capture the dynamic nature of the mediation effect. Specifically, we model the path parameters of mediation models as auto-regressive (AR) processes of time that can vary over time. Additionally, we define the mediation effect under the potential outcome framework, and examine its identification and causal interpretation. Bayesian methods utilizing Gibbs sampling are adopted to estimate unknown parameters in the proposed dynamic mediation models. We further evaluate our proposed models and methods through extensive simulations and illustrate their application through a real data application.

Distributional impacts of anchor borrower program on the welfare of rice farming households in Oyo State, Nigeria

PurposeThis study assessed the heterogeneous impact of the Anchor Borrower Program (ABP) on the welfare distribution of rice farming households in Nigeria.Design/methodology/approachSelf-selection bias and treatment endogeneity were accounted for by employing the Instrumental Variable Quantile Regression (IVQR) model. The estimates obtained from the IVQR model were further compared with those from the conventional quantile regression, and quantile regression using Propensity Score Matching. This was to highlight the extent to which endogeneity bias has been purged from the treatment, in order to establish a consistent causal link between participation in the ABP and the welfare of a cross-section of rice farming households.FindingsABP farmers had significantly higher rice yields across all quantiles of the yield distribution under treatment exogeneity assumption, and in only two quantiles upon controlling for observable confounders. However, this yield gain did not translate to higher Per capita Consumption Expenditure (PCE). The estimates of the more robust IVQR model provided further evidence that the rice yield and PCE of ABP farmers are not statistically different from that of non-ABP farmers across all quantiles of the welfare distribution.Social implicationsThe negligible impact of ABP was relatively higher for lower-yielding households. Thus, implying that, although the ABP is a pro-poor development intervention, the program has not been sufficiently implemented to significantly improve the welfare of the dominant resource-poor farming households in Nigeria.Originality/valueThis study assessed the impact of ABP beyond the conventional potential mean outcome framework by accounting for heterogeneity in treatment effect.Peer reviewThe peer review history for this article is available at: https://publons.com/publon/10.1108/IJSE-02-2023-0083

The policy is always greener: impact heterogeneity of Covid-19 vaccination lotteries in the US.

Covid-19 vaccination has posed crucial challenges to policymakers and health administrations worldwide. Besides the pressure posed by the pandemic, government administrations have to strive against vaccine hesitancy, which seems to be higher with respect to previous vaccination rollouts. To increase the vaccinated population, Ohio announced a monetary incentive as a lottery for those who were vaccinated. 18 other states followed this first example, with varying results. In this paper, we want to evaluate the effect of such policies within the potential outcome framework using the penalized synthetic control method. In the context of staggered treatment adoption, we estimate the effects at a disaggregated level using a panel dataset. We focused on policy outcomes at the county, state, and supra-state levels, highlighting differences between counties with different social characteristics and time frames for policy introduction. We also studied the treatment effect to see whether the impact of these monetary incentives was permanent or only temporary, accelerating the vaccination of citizens who would have been vaccinated in any case.

Weather2vec: Representation Learning for Causal Inference with Non-local Confounding in Air Pollution and Climate Studies

Estimating the causal effects of a spatially-varying intervention on a spatially-varying outcome may be subject to non-local confounding (NLC), a phenomenon that can bias estimates when the treatments and outcomes of a given unit are dictated in part by the covariates of other nearby units. In particular, NLC is a challenge for evaluating the effects of environmental policies and climate events on health-related outcomes such as air pollution exposure. This paper first formalizes NLC using the potential outcomes framework, providing a comparison with the related phenomenon of causal interference. Then, it proposes a broadly applicable framework, termed weather2vec, that uses the theory of balancing scores to learn representations of non-local information into a scalar or vector defined for each observational unit, which is subsequently used to adjust for confounding in conjunction with causal inference methods. The framework is evaluated in a simulation study and two case studies on air pollution where the weather is an (inherently regional) known confounder.

Estimation of Mediation Effect on Zero-Inflated Microbiome Mediators

The mediation analysis methodology of the cause-and-effect relationship through mediators has been increasingly popular over the past decades. The human microbiome can contribute to the pathogenesis of many complex diseases by mediating disease-leading causal pathways. However, standard mediation analysis is not adequate for microbiome data due to the excessive number of zero values and the over-dispersion in the sequencing reads, which arise for both biological and sampling reasons. To address these unique challenges brought by the zero-inflated mediator, we developed a novel mediation analysis algorithm under the potential-outcome framework to fill this gap. The proposed semiparametric model estimates the mediation effect of the microbiome by decomposing indirect effects into two components according to the zero-inflated distributions. The bootstrap algorithm is utilized to calculate the empirical confidence intervals of the causal effects. We conducted extensive simulation studies to investigate the performance of the proposed weighting-based approach and some model-based alternatives, and our proposed model showed robust performance. The proposed algorithm was implemented in a real human microbiome study of identifying whether some taxa mediate the relationship between LACTIN-V treatment and immune response.

Road safety evaluation with multiple treatments: A comparison of methods based on simulations

This paper focuses on ex-post road safety evaluation with multiple treatments. The potential outcome framework for causal inference is introduced to formalize the causal estimands of interest. Various estimation methods are compared via performing simulation experiments based on semi-synthetic data constructed from a London 20 mph zones dataset. The methods under evaluation include regressions, propensity score (PS) based methods, and a machine learning-based method termed generalized random forests (GRF). Both PS-based methods and GRF show higher flexibility with respect to functional specifications of outcome models. Moreover, GRF shows great superiority in the cases where road safety treatments are assigned following specific criteria and/or where there are heterogeneous treatment effects. Considering the ex-post evaluation of combined effects of multiple treatments has significant practical value, the potential outcome framework and the estimation methods presented in this paper are highly recommended for road safety studies.

Rule ensemble method with adaptive group lasso for heterogeneous treatment effect estimation.

The increasing scientific attention given to precision medicine based on real-world data has led to many recent studies clarifying the relationships between treatment effects and patient characteristics. However, this is challenging because of ubiquitous heterogeneity in the treatment effect for individuals and the real-world data on their backgrounds being complex and noisy. Because of their flexibility, various machine learning (ML) methods have been proposed for estimating heterogeneous treatment effect (HTE). However, most ML methods incorporate black-box models that hamper direct interpretation of the relationships between an individual's characteristics and treatment effects. This study proposes an ML method for estimating HTE based on the rule ensemble method RuleFit. The main advantages of RuleFit are interpretability and accuracy. However, HTEs are always defined in the potential outcome framework, and RuleFit cannot be applied directly. Thus, we modified RuleFit and proposed a method to estimate HTEs that directly interpret the relationships among the individuals' features from the model. Actual data from an HIV study, the ACTG 175 dataset, was used to illustrate the interpretation based on the ensemble of rules created by the proposed method. The numerical results confirm that the proposed method has high prediction accuracy compared to previous methods, indicating that the proposed method establishes an interpretable model with sufficient prediction accuracy.

SA6 The Utility of “Single Exposure” Versus Causal Architecture Approaches in Real-World Research

Over recent decades, epidemiologic methodology has evolved substantially to encompass a gamut of techniques for focalizing on the most unbiased effect estimate of a single exposure (e.g., a treatment effect). By contrast, less attention has been paid to the refinement of study designs and analytic techniques intended to encompass the underlying milieu of co-occurring clinical, environmental, and social determinants that interactively influence disease risk or outcomes. Single exposure approaches (i.e., the “potential outcomes framework”) have been criticized as being reductionist and have long been debated in the broader epidemiologic community, despite the utility of this approach in optimizing internal validity. Further, the gap between clinical trial efficacy and real-world (RW) effectiveness remains a concern for both regulatory entities and payers, though this gap may be partly attributable to varying prevalences of important effect modifiers between trial and RW patient populations. This conceptual paper will expand upon the existing conceptual paradigms by applying them to RW research and delineating key considerations related to assessing whether a single exposure or causal architecture approach would be fit-for-purpose based on the overarching research question. Drawing upon examples from RW research types, such as use of external control groups, comparative effectiveness, and prediction of likely responders to treatment, the strengths and weaknesses of each approach will be demonstrated using case studies related to regulatory and/or payer decision-making. Methodologic recommendations will be provided. Ultimately, both single exposure and causal architecture approaches to study design and analysis have a place in pharmacoepidemiology and real-world research. A careful examination of when each approach is fit-for-purpose can lead to the application of innovative strategies to the design and conduct of RW studies.

Long‐term exposure to fine particulate matter is causally linked to increased risk of incident dementia among the American older adults: A national cohort study

AbstractBackgroundGrowing studies link long‐term fine particle (PM2.5) exposures to increased risk of Alzheimer’s disease and related dementias (ADRD), but these findings have been disputed citing traditional approaches do not guarantee evidence of causality.MethodWe constructed two national U.S. population‐based cohorts of enrollees aged ≥65 from the Medicare Chronic Conditions Warehouse (2000‐2018), combined with high‐resolution air pollution datasets, to estimate the effect of long‐term exposure to PM2.5 on dementia and AD incidence, respectively. We applied multiple distinct causal inference approaches, i.e., under a set of causal inference assumptions, to compare the health estimates. We first considered three causal inference modelling approaches that rely on the potential outcomes framework and generalized propensity scores (GPS). These approaches adjust for measured confounding using 1) matching by GPS; 2) weighting by GPS; and 3) including GPS as a covariate in the health outcome model. Then we further applied the difference‐in‐differences approach that controls for unmeasured confounders by design to add robustness to any causal conclusions.ResultFor the period 2000‐2018, we found that all statistical approaches provide consistent results, suggesting that long‐term exposure to PM2.5 is causally linked to increased incidence of ADRD.ConclusionUsing a large nationwide cohort and robust epidemiologic analyses, we provide strong evidence of the causal link between long‐term PM2.5 exposure and increased incidence of dementia and AD. Our results support the importance of continued assessment the impact of air pollution exposure on neurological disorders among the older adults.

Causal mediation and sensitivity analysis for mixed-scale data.

The goal of causal mediation analysis, often described within the potential outcomes framework, is to decompose the effect of an exposure on an outcome of interest along different causal pathways. Using the assumption of sequential ignorability to attain non-parametric identification, Imai et al. (2010) proposed a flexible approach to measuring mediation effects, focusing on parametric and semiparametric normal/Bernoulli models for the outcome and mediator. Less attention has been paid to the case where the outcome and/or mediator model are mixed-scale, ordinal, or otherwise fall outside the normal/Bernoulli setting. We develop a simple, but flexible, parametric modeling framework to accommodate the common situation where the responses are mixed continuous and binary, and, apply it to a zero-one inflated beta model for the outcome and mediator. Applying our proposed methods to the publicly-available JOBS II dataset, we (i) argue for the need for non-normal models, (ii) show how to estimate both average and quantile mediation effects for boundary-censored data, and (iii) show how to conduct a meaningful sensitivity analysis by introducing unidentified, scientifically meaningful, sensitivity parameters.

Associations between environmental noise and myocardial infarction and stroke: Investigating the potential mediating effects of hypertension

BackgroundWe investigated whether hypertension may be a mediator in the pathway linking environmental noise exposure to incident MI and stroke. MethodsSeparately for MI and stroke, we built two population-based cohorts from linked health administrative data. Participants were residents of Montreal (Canada) between 2000 and 2014, aged 45 years and older who were free of hypertension and MI or stroke at time of entry. MI, stroke and hypertension were ascertained from validated case definitions. Residential long-term environmental noise exposure, expressed as the annual mean level acoustic equivalent 24 h (LAeq24h), was estimated from a land use regression model. We performed mediation analysis based on the potential outcomes framework. We used a Cox proportional hazards model for the exposure-outcome model and a logistic regression for the exposure-mediator model. In sensitivity analysis we applied a marginal structural approach to estimate the natural direct and indirect effects. ResultsEach cohort included approximately 900 000 individuals, with 26 647 incident cases of MI and 16 656 incident cases of stroke. 36% of incident MI and 40% of incident stokes had previously developed hypertension. The estimated total effect per interquartile range increase (from 55.0 to 60.5 dB A) in the annual mean LAeq24h was 1.073 (95% confidence interval (CI): 1.070–1.077) for both MI for stroke. We found no evidence of exposure-mediator interaction for both outcomes. The relationships between environmental noise and MI and stroke was not mediated by hypertension. ConclusionsThis population-based cohort study suggests that the main route by which environmental noise exposure may cause MI or stroke is not through hypertension.

A multistate approach for the study of interventions on an intermediate time-to-event in health disparities research.

We propose a novel methodology to quantify the effect of stochastic interventions for a non-terminal intermediate time-to-event on a terminal time-to-event outcome. Investigating these effects is particularly important in health disparities research when we seek to quantify inequities in the timely delivery of treatment and its impact on patients' survival time. Current approaches fail to account for time-to-event intermediates and semi-competing risks arising in this setting. Under the potential outcome framework, we define causal contrasts relevant in health disparities research and provide identifiability conditions when stochastic interventions on an intermediate non-terminal time-to-event are of interest. Causal contrasts are estimated in continuous time within a multistate modeling framework and analytic formulae for the estimators of the causal contrasts are developed. We show via simulations that ignoring censoring in intermediate and/or terminal time-to-event processes or ignoring semi-competing risks may give misleading results. This work demonstrates that a rigorous definition of the causal effects and joint estimation of the terminal outcome and intermediate non-terminal time-to-event distributions are crucial for valid investigation of interventions and mechanisms in continuous time. We employ this novel methodology to investigate the role of delaying treatment uptake in explaining racial disparities in cancer survival in a cohort study of colon cancer patients.

Understanding differences between what alternate propensity score methods estimate.

BACKGROUND: Many approaches to propensity score methods are used in the applied health economics and outcomes research literature. Often this creates confusion when different approaches produce different results for the same data. OBJECTIVE: To present a conceptual overview based on a potential outcomes framework to demonstrate how more than 1 mean treatment effect parameter can be estimated using the propensity score methods and how the selection of appropriate methods should align with the scientific questions. METHODS: We highlight that more than 1 mean treatment effect parameter can be estimated using the propensity score methods. Using the potential outcomes framework and alternate data-generating processes, we discuss under what assumptions different mean treatment effect parameter estimates are supposed to vary. We tie these discussions with propensity score methods to show that different approaches may estimate different parameters. We illustrate these methods using a case study of the comparative effectiveness of apixaban vs warfarin on the likelihood of stroke among patients with a prior diagnosis of atrial fibrillation. RESULTS: Different mean treatment effect parameters take on different values when treatment effects are heterogeneous. We show that traditional propensity score approaches, such as blocking, weighting, matching, or doubly robust, can estimate different mean treatment effect parameters. Therefore, they may not produce the same results even when applied to the same data using the same covariates. We found significant differences in our case study estimates of mean treatment effect parameters. Still, once a mean treatment effect parameter is targeted, estimates across different methods are not different. This highlights the importance of first selecting the target parameter for analysis by aligning the interpretation of the target parameter with the scientific questions and then selecting the specific method to estimate this target parameter. CONCLUSIONS: We present a conceptual overview of propensity score methods in health economics and outcomes research from a potential outcomes framework. We hope these discussions will help applied researchers choose appropriate propensity score approaches for their analysis. DISCLOSURES: Dr Unuigbe's time was supported through an unrestricted postdoctoral fellowship from Pfizer to the University of Washington, Seattle.

Bayesian causal inference: a critical review.

This paper provides a critical review of the Bayesian perspective of causal inference based on the potential outcomes framework. We review the causal estimands, assignment mechanism, the general structure of Bayesian inference of causal effects and sensitivity analysis. We highlight issues that are unique to Bayesian causal inference, including the role of the propensity score, the definition of identifiability, the choice of priors in both low- and high-dimensional regimes. We point out the central role of covariate overlap and more generally the design stage in Bayesian causal inference. We extend the discussion to two complex assignment mechanisms: instrumental variable and time-varying treatments. We identify the strengths and weaknesses of the Bayesian approach to causal inference. Throughout, we illustrate the key concepts via examples. This article is part of the theme issue 'Bayesian inference: challenges, perspectives, and prospects'.

Causal inference and data fusion in econometrics

Summary Learning about cause and effect is arguably the main goal in applied econometrics. In practice, the validity of these causal inferences is contingent on a number of critical assumptions regarding the type of data that has been collected, and the substantive knowledge that is available about the phenomenon under investigation. For instance, unobserved confounding factors threaten the internal validity of estimates; data availability is often limited to nonrandom, selection-biased samples; causal effects need to be learned from surrogate experiments with imperfect compliance; and causal knowledge has to be extrapolated across structurally heterogeneous populations. A powerful and flexible causal inference framework is required in order to tackle all of these challenges, which plague essentially any data analysis to varying degrees. Building on the structural perspective on causality introduced by Haavelmo (1943) and the graph-theoretic approach proposed by Pearl (1995), the artificial intelligence (AI) literature has developed a wide array of techniques for causal inference that allow us to leverage information from various imperfect, heterogeneous, and biased data sources (Bareinboim and Pearl, 2016). In this paper, we review recent advances made in this literature that have the potential to contribute to econometric methodology along three broad dimensions. First, they provide a unified and comprehensive framework for causal learning, in which the above-mentioned problems can be addressed in generality. Second, due to their origin in AI, they come together with sound, efficient, and complete (to be formally defined) algorithmic criteria for automation of the corresponding identification task. And third, because of the nonparametric description of structural models that graph-theoretic approaches build on, they combine the analytical rigor of structural econometrics with the flexibility of the potential outcomes framework, and thus offer a valuable complement to these two literature streams.

Claim causality with clarity.

Abstract Causal inference has attracted enormous attention in clinical neuroscience over the past years. The correct communication and interpretation of causal claims is crucial for the researchers. In this review, we briefly introduced the statistical basis for causal inference in terms of the potential outcome framework and distinguished the different interpretations in experimental and observational studies. Based on the recent efforts in formulating neuroscience research into a causal framework, we provide four recommendations to facilitate better communication of open and reproducible causal inference research.

Invited Perspective: The Potential of Potential Outcomes in Air Pollution Epidemiology.

Invited Perspective: The Potential of Potential Outcomes in Air Pollution Epidemiology.