11. Identification and Estimation of Causal Effects with Time-Varying Treatments and Time-Varying Outcomes

We develop an approach to identifying and estimating causal effects in longitudinal settings with time-varying treatments and time-varying outcomes. The classic potential outcome approach to causal inference generally involves two time periods: units of analysis are exposed to one of two possible values of the causal variable, treatment or control, at a given point in time, and values for an outcome are assessed some time subsequent to exposure. In this paper, we develop a potential outcome approach for longitudinal situations in which both exposure to treatment and the effects of treatment are time-varying. In this longitudinal setting, the research interest centers not on only two potential outcomes, but on a whole matrix of potential outcomes, requiring a complicated conceptualization of many potential counterfactuals. Motivated by sociological applications, we develop a simplification scheme—a weighted composite causal effect that allows identification and estimation of effects with a number of possible solutions. Our approach is illustrated via an analysis of the effects of disability on subsequent employment status using panel data from the Wisconsin Longitudinal Study.

Download Full-text

Causal inference under over-simplified longitudinal causal models

The International Journal of Biostatistics ◽

10.1515/ijb-2020-0081 ◽

2021 ◽

Vol 0 (0) ◽

Author(s):

Lola Étiévant ◽

Vivian Viallon

Keyword(s):

Causal Inference ◽

Causal Effect ◽

Sufficient Conditions ◽

Repeated Measurements ◽

Causal Models ◽

Causal Effects ◽

Sensitivity Analyses ◽

Time Varying ◽

Weighted Averages

Abstract Many causal models of interest in epidemiology involve longitudinal exposures, confounders and mediators. However, repeated measurements are not always available or used in practice, leading analysts to overlook the time-varying nature of exposures and work under over-simplified causal models. Our objective is to assess whether – and how – causal effects identified under such misspecified causal models relates to true causal effects of interest. We derive sufficient conditions ensuring that the quantities estimated in practice under over-simplified causal models can be expressed as weighted averages of longitudinal causal effects of interest. Unsurprisingly, these sufficient conditions are very restrictive, and our results state that the quantities estimated in practice should be interpreted with caution in general, as they usually do not relate to any longitudinal causal effect of interest. Our simulations further illustrate that the bias between the quantities estimated in practice and the weighted averages of longitudinal causal effects of interest can be substantial. Overall, our results confirm the need for repeated measurements to conduct proper analyses and/or the development of sensitivity analyses when they are not available.

Download Full-text

Estimation of causal effects of a time-varying exposure at multiple time points through Multivariable Mendelian randomization

10.1101/2022.01.04.22268740 ◽

2022 ◽

Author(s):

Eleanor Sanderson ◽

Tom G Richardson ◽

Tim T Morris ◽

Kate Tilling ◽

George Davey Smith

Keyword(s):

Direct Effect ◽

Genetic Variants ◽

Causal Effect ◽

Smoking Behaviour ◽

Causal Effects ◽

Mendelian Randomisation ◽

Multiple Time ◽

Time Points ◽

Time Periods ◽

Multiple Time Points

Mendelian Randomisation (MR) is a powerful tool in epidemiology to estimate the causal effect of an exposure on an outcome in the presence of unobserved confounding, by utilising genetic variants as instrumental variables (IVs) for the exposure. The effects obtained from MR studies are often interpreted as the lifetime effect of the exposure in question. However, the causal effects of many exposures are thought to vary throughout an individual's lifetime and there may be periods during which an exposure has more of an effect on a particular outcome. Multivariable MR (MVMR) is an extension of MR that allows for multiple, potentially highly related, exposures to be included in an MR estimation. MVMR estimates the direct effect of each exposure on the outcome conditional on all of the other exposures included in the estimation. We explore the use of MVMR to estimate the direct effect of a single exposure at different time points in an individual's lifetime on an outcome. We use simulations to illustrate the interpretation of the results from such analyses and the key assumptions required. We show that causal effects at different time periods can be estimated through MVMR when the association between the genetic variants used as instruments and the exposure measured at those time periods varies, however this estimation will not necessarily identify exact time periods over which an exposure has the most effect on the outcome. We illustrate the method through estimation of the causal effects of childhood and adult BMI on smoking behaviour.

Download Full-text

Gformula: Estimating Causal Effects in the Presence of Time-Varying Confounding or Mediation using the G-Computation Formula

The Stata Journal Promoting communications on statistics and Stata ◽

10.1177/1536867x1201100401 ◽

2011 ◽

Vol 11 (4) ◽

pp. 479-517 ◽

Cited By ~ 61

Author(s):

Rhian M. Daniel ◽

Bianca L. De Stavola ◽

Simon N. Cousens

Keyword(s):

Indirect Effects ◽

Treatment Effects on Ordinal Outcomes: Causal Estimands and Sharp Bounds

Journal of Educational and Behavioral Statistics ◽

10.3102/1076998618776435 ◽

2018 ◽

Vol 43 (5) ◽

pp. 540-567 ◽

Cited By ~ 3

Author(s):

Jiannan Lu ◽

Peng Ding ◽

Tirthankar Dasgupta

Keyword(s):

Causal Effect ◽

Real Life ◽

Causal Effects ◽

Potential Outcomes ◽

Randomized Experiments ◽

Sharp Bounds ◽

Average Causal Effect ◽

Behavioral Studies ◽

And Control ◽

Potential Outcomes Framework

Assessing the causal effects of interventions on ordinal outcomes is an important objective of many educational and behavioral studies. Under the potential outcomes framework, we can define causal effects as comparisons between the potential outcomes under treatment and control. However, unfortunately, the average causal effect, often the parameter of interest, is difficult to interpret for ordinal outcomes. To address this challenge, we propose to use two causal parameters, which are defined as the probabilities that the treatment is beneficial and strictly beneficial for the experimental units. However, although well-defined for any outcomes and of particular interest for ordinal outcomes, the two aforementioned parameters depend on the association between the potential outcomes and are therefore not identifiable from the observed data without additional assumptions. Echoing recent advances in the econometrics and biostatistics literature, we present the sharp bounds of the aforementioned causal parameters for ordinal outcomes, under fixed marginal distributions of the potential outcomes. Because the causal estimands and their corresponding sharp bounds are based on the potential outcomes themselves, the proposed framework can be flexibly incorporated into any chosen models of the potential outcomes and is directly applicable to randomized experiments, unconfounded observational studies, and randomized experiments with noncompliance. We illustrate our methodology via numerical examples and three real-life applications related to educational and behavioral research.

Download Full-text

Effects of increased body mass index on employment status: a Mendelian randomisation study

International Journal of Obesity ◽

10.1038/s41366-021-00846-x ◽

2021 ◽

Author(s):

Desmond D. Campbell ◽

Michael Green ◽

Neil Davies ◽

Evangelia Demou ◽

Joey Ward ◽

...

Keyword(s):

Body Mass Index ◽

Educational Attainment ◽

Household Income ◽

Employment Status ◽

Early Retirement ◽

Causal Effect ◽

Causal Effects ◽

Mendelian Randomisation ◽

Social Discrimination ◽

Obesity Epidemic

Abstract Background The obesity epidemic may have substantial implications for the global workforce, including causal effects on employment, but clear evidence is lacking. Obesity may prevent people from being in paid work through poor health or through social discrimination. We studied genetic variants robustly associated with body mass index (BMI) to investigate its causal effects on employment. Dataset/methods White UK ethnicity participants of working age (men 40–64 years, women 40–59 years), with suitable genetic data were selected in the UK Biobank study (N = 230,791). Employment status was categorised in two ways: first, contrasting being in paid employment with any other status; and second, contrasting being in paid employment with sickness/disability, unemployment, early retirement and caring for home/family. Socioeconomic indicators also investigated were hours worked, household income, educational attainment and Townsend deprivation index (TDI). We conducted observational and two-sample Mendelian randomisation (MR) analyses to investigate the effect of increased BMI on employment-related outcomes. Results Regressions showed BMI associated with all the employment-related outcomes investigated. MR analyses provided evidence for higher BMI causing increased risk of sickness/disability (OR 1.08, 95% CI 1.04, 1.11, per 1 Kg/m2 BMI increase) and decreased caring for home/family (OR 0.96, 95% CI 0.93, 0.99), higher TDI (Beta 0.038, 95% CI 0.018, 0.059), and lower household income (OR 0.98, 95% CI 0.96, 0.99). In contrast, MR provided evidence for no causal effect of BMI on unemployment, early retirement, non-employment, hours worked or educational attainment. There was little evidence for causal effects differing by sex or age. Robustness tests yielded consistent results. Discussion BMI appears to exert a causal effect on employment status, largely by affecting an individual’s health rather than through increased unemployment arising from social discrimination. The obesity epidemic may be contributing to increased worklessness and therefore could impose a substantial societal burden.

Download Full-text

Distinguishing genetic correlation from causation across 52 diseases and complex traits

10.1101/205435 ◽

2017 ◽

Cited By ~ 11

Author(s):

Luke J. O’Connor ◽

Alkes L. Price

Keyword(s):

Genetic Correlation ◽

Complex Traits ◽

Causal Effect ◽

Genetic Correlations ◽

Genetic Component ◽

Causal Effects ◽

Marginal Effect ◽

Mineral Density ◽

Wide Range ◽

Causal Variable

AbstractMendelian randomization (MR) is widely used to identify causal relationships among heritable traits, but it can be confounded by genetic correlations reflecting shared etiology. We propose a model in which a latent causal variable mediates the genetic correlation between two traits. Under the latent causal variable (LCV) model, trait 1 is fully genetically causal for trait 2 if it is perfectly genetically correlated with the latent causal variable, implying that the entire genetic component of trait 1 is causal for trait 2; it is partially genetically causal for trait 2 if it has a high genetic correlation with the latent variable, implying that part of the genetic component of trait 1 is causal for trait 2. To quantify the degree of partial genetic causality, we define the genetic causality proportion (gcp). We fit this model using mixed fourth moments E(α1α2) and E(α1α2) of marginal effect sizes for each trait, exploiting the fact that if trait 1 is causal for trait 2 then SNPs affecting trait 1 (large ) will have correlated effects on trait 2 (large α1α2), but not vice versa. We performed simulations under a wide range of genetic architectures and determined that LCV, unlike state-of-the-art MR methods, produced well-calibrated false positive rates and reliable gcp estimates in the presence of genetic correlations and asymmetric genetic architectures; we also determined that LCV is well-powered to detect a causal effect. We applied LCV to GWAS summary statistics for 52 traits (average N=331k), identifying partially or fully genetically causal effects (1% FDR) for 59 pairs of traits, including 30 pairs of traits with high gcp estimates (gĉp > 0.6). Results consistent with the published literature included genetically causal effects on myocardial infarction (MI) for LDL, triglycerides and BMI. Novel findings included a genetically causal effect of LDL on bone mineral density, consistent with clinical trials of statins in osteoporosis. These results demonstrate that it is possible to distinguish between genetic correlation and causation using genetic data.

Download Full-text

Bayesian Inference of Causal Effects for an Ordinal Outcome in Randomized Trials

Journal of Causal Inference ◽

10.1515/jci-2017-0019 ◽

2018 ◽

Vol 6 (2) ◽

Cited By ~ 2

Author(s):

Yasutaka Chiba

Keyword(s):

Randomized Trials ◽

Randomized Clinical Trials ◽

Likelihood Function ◽

Causal Effect ◽

Causal Effects ◽

Potential Outcomes ◽

Response Type ◽

Binary Outcome ◽

Modeling Assumption ◽

Ordinal Outcome

AbstractIn randomized trials in which two treatment arms are compared with a binary outcome, the causal effect can be identified by assuming that the two treatment arms are exchangeable. In trials with an ordinal outcome, which is categorized as more than two, the causal effect can be identified by assuming that the potential outcomes are independent and that the two treatment arms are exchangeable. In this article, we propose a Bayesian approach to causal inference that does not rely on these two assumptions. To achieve this purpose, we use a randomization-based approach and response type. Then, the likelihood function is derived by physical randomization in which subjects who belong to a response type are randomly assigned to the treatment or control, with no modeling assumption on the outcome. Our approach can derive not only the posterior distribution of the causal effect but also that of the number of subjects in each response type. The proposed approach is illustrated with two examples from randomized clinical trials.

Download Full-text

G-estimation of causal pathways in vocational rehabilitation for adults with psychotic disorders – a secondary analysis of a randomized trial

BMC Psychiatry ◽

10.1186/s12888-021-03349-1 ◽

2021 ◽

Vol 21 (1) ◽

Author(s):

Ole Klungsøyr ◽

June Ullevoldsæter Lystad ◽

Helen Bull ◽

Stig Evensen ◽

Torill Ueland ◽

...

Keyword(s):

Vocational Rehabilitation ◽

Negative Symptoms ◽

Causal Effect ◽

Working Hours ◽

Causal Effects ◽

Psychotic Symptoms ◽

Time Varying ◽

Occupational Functioning ◽

Data Set ◽

Causal Pathways

Abstract Background Vocational rehabilitation (VR) has increasingly become an important intervention targeting poor occupational functioning in schizophrenia. The Norwegian Job Management Program (JUMP), sought to enhance occupational outcomes by augmenting VR with either cognitive behavioral therapy (CBT) techniques aiming to improve psychotic symptoms or cognitive remediation (CR) aiming to improve cognition. CBT is standard treatment in schizophrenia, but recent meta-analyses question the effect of CBT on negative psychotic symptoms. It is of interest to study the causal role of psychotic symptoms and cognitive functioning on occupational functioning. Methods Data from the JUMP VR – program, was reanalyzed with a causal inference method to assess the causal effects of reduced symptoms / improved neurocognitive functioning on occupational functioning measured by number of working hours per week. Participants (N = 131) had been randomized to either VR + CBT (N = 68) or VR + CR (N = 63). Large improvements in number of working hours were demonstrated in both intervention groups (nonsignificant group difference). G-estimation was used to assess the strength and nature of the causal effects, adjusted for time-varying confounding and selection – bias from loss to follow-up. Results Significant causal effects of reduction in each of four dimensions of symptoms and improved neurocognition respectively, on number of working hours were found (separate models). The effect of negative symptoms was the strongest and increased in magnitude during the whole observation period, while the effect of two other symptoms and neurocognition was constant. Adjusted for confounding (including potential feedback), the causal effect of a hypothetical change in negative symptoms equal to the average improvement in the CBT group corresponded to an increase in working hours of 3.2 h per week (95% CI: 1.11, 5.35). Conclusion High performance of g-estimation in a small psychiatric data set with few repeated measures and time-varying confounding and effects, was demonstrated. Augmented vocational rehabilitation showed causal effects of intervention targets with the strongest and increasing effect from negative symptoms on number of working hours. Combination of therapy and activation (indirect and direct approach) might explain improvement in both cognition and negative symptoms, and shed some light on effective ingredients for improved treatment of negative symptoms.

Download Full-text

G-Estimation of Causal Pathways in Vocational Rehabilitation For Adults With Psychotic Disorders – A Secondary Analysis of A Randomized Trial

10.21203/rs.3.rs-279135/v1 ◽

2021 ◽

Author(s):

Ole Klungsøyr ◽

June Ullevoldsæter Lystad ◽

Helen Bull ◽

Stig Evensen ◽

Torill Ueland ◽

...

Keyword(s):

Vocational Rehabilitation ◽

Negative Symptoms ◽

Causal Effect ◽

Working Hours ◽

Causal Effects ◽

Psychotic Symptoms ◽

Time Varying ◽

Occupational Functioning ◽

Data Set ◽

Causal Pathways

Abstract Background: Vocational rehabilitation (VR) has increasingly become an important intervention targeting poor occupational functioning in schizophrenia. The Norwegian Job Management Program (JUMP), sought to enhance occupational outcomes by augmenting VR with either cognitive behavioral therapy (CBT) techniques aiming to improve psychotic symptoms or cognitive remediation (CR) aiming to improve cognition. CBT is standard treatment in schizophrenia, but recent meta-analyses question the effect of CBT on negative psychotic symptoms. It is of interest to study the causal role of psychotic symptoms and cognitive functioning on occupational functioning. Methods: Data from the JUMP VR – program, was reanalyzed with a causal inference method to assess the causal effects of reduced symptoms / improved neurocognitive functioning on occupational functioning measured by number of working hours per week. Participants (N=131) had been randomized to either VR + CBT (N=68) or VR + CR (N=63). Large improvements in number of working hours were demonstrated in both intervention groups (nonsignificant group difference). G-estimation was used to assess the strength and nature of the causal effects, adjusted for time-varying confounding and selection – bias from loss to follow-up. Results: Significant causal effects of reduction in each of four dimensions of symptoms and improved neurocognition respectively, on number of working hours were found (separate models). The effect of negative symptoms was the strongest and increased in magnitude during the whole observation period, while the effect of the other symptoms and neurocognition was constant. Adjusted for confounding (including potential feedback), the causal effect of a hypothetical change in negative symptoms equal to the average improvement in the CBT group corresponded to an increase in working hours of 3.2 hours per week (95% CI: 1.11, 5.35).Conclusion: High performance of g-estimation in a small psychiatric data set with few repeated measures and time-varying confounding and effects, was demonstrated. Augmented vocational rehabilitation showed causal effects of intervention targets with the strongest and increasing effect from negative symptoms on number of working hours.Combination of therapy and activation (indirect and direct approach) might explain improvement in both cognition and negative symptoms, and shed some light on effective ingredients for improved treatment of negative symptoms.

Download Full-text

Understanding Distressed Residential Transaction Discounts Across Price Segments and Market Conditions

Journal of Real Estate Research ◽

10.22300/0896-5803.42.1.151 ◽

2020 ◽

Vol 42 (1) ◽

pp. 151-182

Author(s):

Ramya Rajajagadeesan Aroul ◽

J. Andrew Hansz ◽

Mauricio Rodriguez

Keyword(s):

Transaction Costs ◽

Time Varying ◽

Market Segment ◽

Market Conditions ◽

Large Market ◽

Space And Time ◽

Wide Range ◽

Time Periods ◽

Market Areas

In the literature, there is a wide range of discounts associated with foreclosures. Comparisons across studies are difficult as they use different methodologies across large areas over different time periods. We employ a consistent methodology across space and time. We find modest discounts, within the range of typical transaction costs, in all but the highest priced market segment. Higher priced segments could explain prior findings of substantial discounts. We find that discounts are time-varying, with discounts increasing with market distress. A one-size-fits-all approach is not appropriate when estimating distressed transaction discounts across large market areas or under changing market conditions.

Download Full-text