Separable Effects for Causal Inference in the Presence of Competing Events

Stensrud, Mats Julius; Young, Jessica; Didelez, Vanessa; Robins, James M.; Hernán, Miguel A.

doi:10.1080/01621459.2020.1765783

Cited by 67 publications

(134 citation statements)

References 18 publications

Supporting

Mentioning

130

Contrasting

Order By: Relevance

“…While IPCW is a useful tool to advance causal inference, the interpretation of the estimated effects after IPCW may not be straightforward in the presence of competing risk. [35][36][37] Moreover, the weight calculation requires the information on exposure and covariates among the censored individuals, which often is unavailable. 38…”

Section: Ipw Can Be Used To Address Selection Bias Toomentioning

confidence: 99%

Using Propensity Scores for Causal Inference: Pitfalls and Tips

Shiba

Kawahara

2021

Journal of Epidemiology

View full text Add to dashboard Cite

show abstract

Section: Ipw Can Be Used To Address Selection Bias Toomentioning

confidence: 99%

Using Propensity Scores for Causal Inference: Pitfalls and Tips

Shiba

Kawahara

2021

Journal of Epidemiology

View full text Add to dashboard Cite

show abstract

“…One can also consider a combined outcome endpoint, such as the effect on dementia or death, but this too is unsatisfactory in many cases: in our example, the effect of smoking on risk of death would overwhelm. A novel alternative, the so-called separable effects avoid evoking consideration of implausible scenarios that "eliminate death" or unobservable subpopulations [21]. The separable effects are effects of modified treatments that are assumed to operate like the study treatment but with particular mechanisms removed.…”

Section: Discussionmentioning

confidence: 99%

“…Instead of a total effect, a direct effect of smoking on the risk of dementia (that does not also capture the pathways mediated by death) may be of interest. There are multiple ways to define a direct effect [20][21][22]. Here we will consider one definition that has been historically considered and may lead to familiar statistical methods as will be described in the next section: the controlled direct effect.…”

Section: Choosing a Causal Question: The Total And Controlled Direct Effectmentioning

confidence: 99%

Choosing questions before methods in dementia research with competing events and causal goals

Young

Didelez

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Several of the hypothesized or studied exposures that may affect dementia risk are known to increase the risk of death. This may explain counterintuitive results, where exposures that are known to be harmful for mortality risk sometimes seem protective for the risk of dementia. Authors have attempted to explain these counterintuitive results as biased, but the bias associated with a particular analytic method cannot be defined or assessed if the causal question is not explicitly specified. Indeed, we can consider several causal questions when competing events like death, which cannot be prevented by design, are present. Current dementia research guidelines have not explicitly considered what constitutes a meaningful causal question in this setting or, more generally, how this choice justifies and should drive particular analytic decisions. To contextualize current practices, we first perform a systematic review of the conduct and interpretation of longitudinal studies focused on dementia outcomes where death is a competing event. We then describe and demonstrate how to address different causal questions (referred here as "the total effect" and "the controlled direct effect") with traditional analytic approaches under explicit assumptions. Our application focuses on smoking cessation in late-midlife. To illustrate core concepts, we discuss this example both in terms of a hypothetical randomized trial and with an emulation of such a trial using observational data from the Rotterdam Study.

show abstract

“…The assumption of consistency is essential for matching methods to resemble those methods based on causal theories supported by the logic of controlled experiments. Consistency requires that the outcome of a given unit, belonging to the treatment group, in a non‐experimental scenario, is similar, if not identical, to the outcome of the same intervention in the same unit in a controlled experimental scenario (Stensrud et al ., 2020). As exemplified by Pearl & Mackenzie (2018, p. 255): “a person who took aspirin and recovered would also recover if given aspirin by experimental design.”…”

Section: Assumptions Of the Potential Outcome Framework For Applications Of Matching Methodsmentioning

confidence: 99%

Estimating counterfactuals for evaluation of ecological and conservation impact: an introduction to matching methods

2021

View full text Add to dashboard Cite

Matching methods encompass non‐parametric approaches to estimating counterfactual states through a rigorous selection of control units with similar characteristics to units submitted to an intervention. These methods enable comparisons between treated and control units in a way that facilitates understanding of causal relationships between interventions and outcomes. Matching methods have been used only recently in ecology and conservation biology, where such applications changed the way the field investigates causal questions, for example, in impact‐evaluation studies. However, the strengths and limitations of matching methods are not well understood by most ecologists and environmental scientists. Herein, we review state‐of‐the‐art matching methods aiming to help fill this gap in understanding. First, we present relevant theoretical concepts related to matching methods and related subjects such as counterfactual states and causation. Next, we propose guidelines and strategies for the application of matching methods in ecology and conservation biology. Finally, we discuss the possibilities for future applications of matching methods in the environmental sciences.

show abstract

Separable Effects for Causal Inference in the Presence of Competing Events

Cited by 67 publications

References 18 publications

Using Propensity Scores for Causal Inference: Pitfalls and Tips

Using Propensity Scores for Causal Inference: Pitfalls and Tips

Choosing questions before methods in dementia research with competing events and causal goals

Estimating counterfactuals for evaluation of ecological and conservation impact: an introduction to matching methods

Contact Info

Product

Resources

About