Inference In Observational Studies Research Articles

Towards the year 2049: The next 25 years of occupational health and safety research.

In this discussion paper, we close our 2024 series reflecting on the successes, failures, and promises of occupational health and safety research in celebration of the 50th anniversary of the Scandinavian Journal of Work, Environment & Health (SJWEH). This paper aims to elaborate on the future of our research field. We conducted a narrative review of lessons learned in the series, examining insights gained and key takeaways. Additionally, we explored the current and anticipated agendas of major institutions, including the World Health Organization and the European Union, on occupational health and safety, as well as potential developments in the academic publishing industry. Occupational health and safety research has significantly evolved over the last 50 years, emphasizing longitudinal study designs, enriching observational data with registry-based information, and expanding the scope of hazardous determinants impacting workers` health. Novel statistical approaches have further enabled researchers to address complex associations, such as mediation effects, and to strengthen causal inference in observational studies. At the same time, the publishing business is changing rapidly, with artificial intelligence poised to reshape both research practices and the landscape of academic publishing. In the changing landscape of research and academic publishing, our goal is for SJWEH to continue to be a leading source of high-quality research dedicated to protecting and improving workers' health. We are curious and excited to see where all these current and anticipated changes will lead in the years to come.

What Explains Hop Growers' Fungicide Use Intensity and Management Costs in Response to Powdery Mildew?

Methods for causal inference from observational data are common in human disease epidemiology and social sciences but are used relatively little in plant pathology. We draw upon an extensive data set of the incidence of hop plants with powdery mildew (caused by Podosphaera macularis) collected from yards in Oregon from 2014 to 2017 and associated metadata on grower cultural practices, cultivar susceptibility to powdery mildew, and pesticide application records to understand variation in and causes of growers' fungicide use and associated costs. An instrumental causal forest model identified growers' spring pruning thoroughness, cultivar susceptibility to two of the dominant pathogenic races of P. macularis, network centrality of yards during May-June and June-July time transitions, and the initial strain of the fungus detected as important variables determining the number of pesticide active constituents applied by growers and the associated costs they incurred in response to powdery mildew. Exposure-response function models fit after covariate weighting indicated that both the number of pesticide active constituents applied and their associated costs scaled linearly with the seasonal mean incidence of plants with powdery mildew. Although the causes of pesticide use intensity are multifaceted, biological and production factors collectively influence the incidence of powdery mildew, which has a direct exposure-response relationship with the number of pesticide active constituents that growers apply and their costs. Our analyses point to several potential strategies for reducing pesticide use and costs for management of powdery mildew on hop. We also highlight the utility of these methods for causal inference in observational studies.

PyTWMR: transcriptome-wide Mendelian randomization in python.

Mendelian randomization (MR) is a widely used approach to estimate causal effect of variation in gene expression on complex traits. Among several MR-based algorithms, transcriptome-wide summary statistics-based Mendelian Randomization approach (TWMR) enables the uses of multiple SNPs as instruments and multiple gene expression traits as exposures to facilitate causal inference in observational studies. Here we present a Python-based implementation of TWMR and revTWMR. Our implementation offers GPU computational support for faster computations and robust computation mode resilient to highly correlated gene expressions and genetic variants. pyTWMR is available at github.com/soreshkov/pyTWMR.

RoA: visual analytics support for deconfounded causal inference in observational studies

The gold standard in medical research to estimate the causal effect of a treatment is the Randomized Controlled Trial (RCT), but in many cases these are not feasible due to ethical, financial or practical issues. Observational studies are an alternative, but can easily lead to doubtful results, because of unbalanced selection bias and confounding. Moreover, RCTs often only apply to a specific subgroup and cannot readily be extrapolated. In response, we present Rod of Asclepius (RoA), a novel visual analytics method that integrates modern techniques designed for identification of causal effects and effect size estimation with subgroup analysis. The result is an interactive display designed to combine exploratory analysis with a robust set of techniques, including causal do-calculus, propensity score weighting, and effect estimation. It enables analysts to conduct observational studies in an exploratory, yet robust way. This is demonstrated by means of a use case involving patients undergoing surgery, for which we collaborated closely with clinical researchers.

A doubly robust estimator for the Mann Whitney Wilcoxon rank sum test when applied for causal inference in observational studies

ABSTRACT The Mann–Whitney–Wilcoxon rank sum test (MWWRST) is a widely used method for comparing two treatment groups in randomized control trials, particularly when dealing with highly skewed data. However, when applied to observational study data, the MWWRST often yields invalid results for causal inference. To address this limitation, Wu et al. (Causal inference for Mann–Whitney–Wilcoxon rank sum and other nonparametric statistics, Stat. Med. 33 (2014), pp. 1261–1271) introduced an approach that incorporates inverse probability weighting (IPW) into this rank-based statistic to mitigate confounding effects. Subsequently, Mao (On causal estimation using U-statistics, Biometrika 105 (2018), pp. 215–220), Zhang et al. (Estimating Mann Whitney-type causal effects, J. Causal Inference 7 (2019), ARTICLE ID 20180010), and Ai et al. (A Mann–Whitney test of distributional effects in a multivalued treatment, J. Stat. Plan. Inference 209 (2020), pp. 85–100) extended this IPW estimator to develop doubly robust estimators. Nevertheless, each of these approaches has notable limitations. Mao's method imposes stringent assumptions that may not align with real-world study data. Zhang et al.'s (Estimating Mann Whitney-type causal effects, J. Causal Inference 7 (2019), ARTICLE ID 20180010) estimators rely on bootstrap inference, which suffers from computational inefficiency and lacks known asymptotic properties. Meanwhile, Ai et al. (A Mann–Whitney test of distributional effects in a multivalued treatment, J. Stat. Plan. Inference 209 (2020), pp. 85–100) primarily focus on testing the null hypothesis of equal distributions between two groups, which is a more stringent assumption that may not be well-suited to the primary practical application of MWWRST. In this paper, we aim to address these limitations by leveraging functional response models (FRM) to develop doubly robust estimators. We demonstrate the performance of our proposed approach using both simulated and real study data.

The application of target trials with longitudinal targeted maximum likelihood estimation to assess the effect of alcohol consumption in adolescence on depressive symptoms in adulthood.

Time-varying confounding is a common challenge for causal inference in observational studies with time-varying treatments, long follow-up periods, and participant dropout. Confounder adjustment using traditional approaches can be limited by data sparsity, weight instability, and computational issues. The Nicotine Dependence in Teens Study is a prospective cohort study, and we used data from 21 data collection cycles carried out from 1999 to 2008 among 1294 students recruited from 10 high schools in Montreal, Quebec, Canada, including follow-up into adulthood. Our aim in this study was to estimate associations of timing of alcohol initiation and cumulative duration of alcohol use with depression symptoms in adulthood. Based on the target trials framework, we defined intention-to-treat and as-treated parameters in a marginal structural model with sex as a potential effect-modifier. We then used the observational data to emulate the trials. For estimation, we used pooled longitudinal target maximum likelihood estimation, a plug-in estimator with double-robust and local efficiency properties. We describe strategies for dealing with high-dimensional potential drinking patterns and practical positivity violations due to a long follow-up time, including modifying the effect of interest by removing sparsely observed drinking patterns from the loss function and applying longitudinal modified treatment policies to represent the effect of discouraging drinking.

Causal effects of later-eating rhythm on adiposity in children through the comparison of two cohorts in the UK and China: a cross-cohort study

BackgroundLater-eating rhythm (LER) refers to a later timing, greater energy intake, and higher meal frequency in the evening. The role of childhood LER in obesity development is emerging, but most evidence is cross-sectional. Cross-context comparison allows the improvement of causal inference in observational studies by comparing cohorts with different confounding structures. This method is applied to assess the causal effects of LER on adiposity, by exploring the likelihood of residual confounding due to socioeconomic status. MethodsIn this cross-cohort analysis, we used ongoing birth cohort data from the UK Avon Longitudinal Study of Parents and Children (ALSPAC) established in 1991, and the nationally representative China Health and Nutrition Survey (CHNS) collected in 1989–2011. Children with available data at age 7 years were eligible. We applied indices of inequality for assessing confounding structure by regressing LER/adiposity on the standardised score of socioeconomic status (SES) in each cohort. We used multivariable linear and binary logistic regressions to model cross-sectional and prospective associations between LER at 7 years of age and body-mass index (BMI) at ages 7 and 9 years in both cohorts. Analyses were adjusted by age, sex, ethnicity, residency, and socioeconomic status. We used a p value for the Cochrane Q-test obtained from meta-analysis to test for heterogeneity between cohorts. FindingsWe analysed data from 4019 children (2170 [54·0%] female; 1849 [46.0%] male) in ALSPAC and 1749 (788 [45·1%] female; 961 [54.9%] male) in CHNS. The associations between SES and LER or adiposity differed between ALSPAC and CHNS (SES and energy intake for evening main meal: b=1·81 [95% CI 0·81 to 2·81] vs –3·02 [–4·76 to –1·27]; SES and frequency of evening snacks: odds ratio [OR]=0·51 [95% CI 0·41–0·63] vs 5·71 [3·54–9·22]; SES and BMI: b=–0·42 [–0·65 to –0·18] vs 1·29 [0·75 to 1·84]). Positive associations between frequency of evening snacks and BMI were seen in both cross-sectional and longitudinal analyses in both cohorts (mean change of BMI with 1 day increase of consuming evening snacks b=0·09 [0·02 to 0·15]; 0·13 [0·03 to 0·22] kg/m2 per day in ALSPAC, and b=0·11 [–0·07 to 0·28]; 0·30 [0·07 to 0·52] kg/m2 per day in CHNS). No associations were found for energy intake. p values for heterogeneity ranged from 0·107 to 0·932. InterpretationBoth cohorts showed consistent results despite varied dietary cultures and SES patterning of LER or adiposity. Energy intake in the evening or night was not associated with adiposity, whereas evening snacking was. More recent, high-quality cohorts are warranted to enhance the strength of the conclusions. FundingNone.

The directed acyclic graph helped identify confounders in the association between coronary heart disease and pesticide exposure among greenhouse vegetable farmers.

To explore the causal pathways associated with coronary heart disease (CHD) and pesticide exposure using a directed acyclic graph (DAG) analysis and to investigate the potential benefits of DAG by comparing it with logistic regression. This cross-sectional study enrolled 1368 participants from April 2015 to May 2017. Trained research investigators interviewed farmers using a self-administered questionnaire. Logistic regression and DAG models were used to identify the associations between CHD and chronic pesticide exposure. A total of 150 (11.0%) of the 1368 participants are characterized as having CHD. High pesticide exposure (odds ratio = 2.852, 95% confidence intervals: 1.951-4.171) is associated with CHD when compare with low pesticide exposure by both DAG and logistic analyses. After adjusting for the additional potential influence of factors identified by the DAG analysis, there is no significant association, such as the results in logistic regression: ethnicity, education level, settlement time, and mixed pesticide status. Specifically, age, meal frequency, and consumption of fresh fruit, according to the DAG analysis, are independent factors for CHD. High pesticide exposure is a risk factor for CHD as indicated by both DAG and logistic regression analyses. DAG can be a preferable improvement over traditional regression methods to identify sources of bias and causal inference in observational studies, especially for complex research questions.

Regression-assisted Bayesian record linkage for causal inference in observational studies with covariates spread over two files

We consider causal inference for observational studies with data spread over two files. One file includes the treatment, outcome, and some covariates measured on a set of individuals, and the other file includes additional causally-relevant covariates measured on a partially overlapping set of individuals. By linking records in the two databases, the analyst can control for more covariates, thereby reducing the risk of bias compared to using only one file alone. When analysts do not have access to a unique identifier that enables perfect, error-free linkages, they typically rely on probabilistic record linkage to construct a single linked data set, and estimate causal effects using these linked data. This typical practice does not propagate uncertainty from imperfect linkages to the causal inferences. Further, it does not take advantage of relationships among the variables to improve the linkage quality. We address these shortcomings by fusing regression-assisted, Bayesian probabilistic record linkage with causal inference. The Markov chain Monte Carlo sampler generates multiple plausible linked data files as byproducts that analysts can use for multiple imputation inferences. Here, we show results for two causal estimators based on propensity score overlap weights. Using simulations and data from the Italy Survey on Household Income and Wealth, we show that our approach can improve the accuracy of estimated treatment effects.

Reconciling Algorithmic Fairness Criteria

Reconciling Algorithmic Fairness Criteria

Complex discontinuity designs using covariates: Impact of school grade retention on later life outcomes in Chile

Regression discontinuity designs are extensively used for causal inference in observational studies. However, they are usually confined to settings with simple treatment rules and determined by a single running variable with a single cutoff. Motivated by the problem of estimating the impact of grade retention on educational and juvenile crime outcomes in Chile, we propose a framework and methods for complex discontinuity designs that encompass multiple treatment rules. In this framework the observed covariates play a central role for identification, estimation, and generalization of causal effects. Identification is nonparametric and relies on a local strong ignorability assumption. Estimation proceeds, as in any observational study, under strong ignorability, yet in a neighborhood of the cutoffs of the running variables. We discuss estimation approaches based on matching and weighting, including complementary regression modeling adjustments. We present assumptions for generalization, that is, for identification and estimation of average treatment effects for target populations. We also describe two approaches to select the neighborhood for analysis. We find that grade retention in Chile has a negative impact on future grade retention but is not associated with dropping out of school or committing a juvenile crime.

Large Sample Properties of Matching for Balance

Matching methods are widely used for causal inference in observational studies. Among them, nearest neighbor matching is arguably the most popular. However, nearest neighbor matching does not generally yield an average treatment effect estimator that is $\sqrt{n}$-consistent (Abadie and Imbens, 2006). Are matching methods not $\sqrt{n}$-consistent in general? In this paper, we study a recent class of matching methods that use integer programming to directly target aggregate covariate balance as opposed to finding close neighbor matches. We show that under suitable conditions these methods can yield simple estimators that are $\sqrt{n}$-consistent and asymptotically optimal.

Lmtp: An R Package for Estimating the Causal Effects of Modified Treatment Policies

We present the lmtp R package for causal inference from longitudinal observational or randomized studies. This package implements the estimators of Díaz et al. (2021) for estimating general non-parametric causal effects based on modified treatment policies. Modified treatment policies generalize static and dynamic interventions, making lmtp and all-purpose package for non-parametric causal inference in observational studies. The methods provided can be applied to both point-treatment and longitudinal settings, and can account for time-varying exposure, covariates, and right censoring thereby providing a very general tool for causal inference. Additionally, two of the provided estimators are based on flexible machine learning regression algorithms, and avoid bias due to parametric model misspecification while maintaining valid statistical inference.

Improving the Validity of Causal Inferences in Observational Studies.

Editor's note: This is the 13th article in a series on clinical research by nurses. The series is designed to be used as a resource for nurses to understand the concepts and principles essential to research. Each column will present the concepts that underpin evidence-based practice-from research design to data interpretation. To see all the articles in the series, go to https://links.lww.com/AJN/A204.

Combining parametric and nonparametric models to estimate treatment effects in observational studies.

Performing causal inference in observational studies requires we assume confounding variables are correctly adjusted for. In settings with few discrete-valued confounders, standard models can be employed. However, as the number of confounders increases these models become less feasible as there are fewer observations available for each unique combination of confounding variables. In this paper, we propose a new model for estimating treatment effects in observational studies that incorporates both parametric and nonparametric outcome models. By conceptually splitting the data, we can combine these models while maintaining a conjugate framework, allowing us to avoid the use of Markov chain Monte Carlo (MCMC) methods. Approximations using the central limit theorem and random sampling allow our method to be scaled to high-dimensional confounders. Through simulation studies we show our method can be competitive with benchmark models while maintaining efficient computation, and illustrate the method on a large epidemiological health survey.

Net-benefit regression with censored cost-effectiveness data from randomized or observational studies.

Cost-effectiveness analysis is an essential part of the evaluation of new medical interventions. While in many studies both costs and effectiveness (eg, survival time) are censored, standard survival analysis techniques are often invalid due to the induced dependent censoring problem. We propose methods for censored cost-effectiveness data using the net-benefit regression framework, which allow covariate-adjustment and subgroup identification when comparing two intervention groups. The methods provide a straightforward way to construct cost-effectiveness acceptability curves with censored data. We also propose a more efficient doubly robust estimator of average causal incremental net benefit, which increases the likelihood that the results will represent a valid inference in observational studies. Lastly, we conduct extensive numerical studies to examine the finite-sample performance of the proposed methods, and illustrate the proposed methods with a real data example using both survival time and quality-adjusted survival time as the measures of effectiveness.

Overview on the generalized propensity scoring estimator for continuous treatment

Among kinds of methods for causal inference in observational studies, the propensity score (PS) method to control measured confounding is more widely used. PS method usually consists of two critical steps: first, estimating the propensity score, followed by calculating the causal parameters of interest by regression, weighting, matching, and stratification. Unlike the traditional dichotomous treatment, the generalized propensity scoring estimator used for continuous treatment has been proposed in recent years. Many methods have been developed to estimate the generalized propensity score or even estimate the balancing weight directly. This paper introduces the existing estimators from both the model-based and balance-based perspectives.

The impact of unmeasured confounding on causal inference in observational studies: A plasmode simulation study of targeted maximum likelihood estimation

The impact of unmeasured confounding on causal inference in observational studies: A plasmode simulation study of targeted maximum likelihood estimation

Application of Standardization for Causal Inference in Observational Studies: A Step-by-step Tutorial for Analysis Using R Software.

Epidemiological studies typically examine the causal effect of exposure on a health outcome. Standardization is one of the most straightforward methods for estimating causal estimands. However, compared to inverse probability weighting, there is a lack of user-centric explanations for implementing standardization to estimate causal estimands. This paper explains the standardization method using basic R functions only and how it is linked to the R package stdReg, which can be used to implement the same procedure. We provide a step-by-step tutorial for estimating causal risk differences, causal risk ratios, and causal odds ratios based on standardization. We also discuss how to carry out subgroup analysis in detail.

Randomization Tests in Observational Studies With Staggered Adoption of Treatment

ABSTRACT This article considers the problem of inference in observational studies with time-varying adoption of treatment. In addition to an unconfoundedness assumption that the potential outcomes are independent of the times at which units adopt treatment conditional on the units’ observed characteristics, our analysis assumes that the time at which each unit adopts treatment follows a Cox proportional hazards model. This assumption permits the time at which each unit adopts treatment to depend on the observed characteristics of the unit, but imposes the restriction that the probability of multiple units adopting treatment at the same time is zero. In this context, we study randomization tests of a null hypothesis that specifies that there is no treatment effect for all units and all time periods in a distributional sense. We first show that an infeasible test that treats the parameters of the Cox model as known has rejection probability under the null hypothesis no greater than the nominal level in finite samples. Since these parameters are unknown in practice, this result motivates a feasible test that replaces these parameters with consistent estimators. While the resulting test does not need to have the same finite-sample validity as the infeasible test, we show that it has limiting rejection probability under the null hypothesis no greater than the nominal level. In a simulation study, we examine the practical relevance of our theoretical results, including robustness to misspecification of the model for the time at which each unit adopts treatment. Finally, we provide an empirical application of our methodology using the synthetic control-based test statistic and tobacco legislation data found in Abadie, Diamond and Hainmueller. Supplementary materials for this article are available online.