Search

Directed acyclic graphs (DAGs) are a useful tool to represent, in a graphical format, researchers’ assumptions about the causal structure among variables while providing a rationale for the choice of confounding variables to adjust for. With origins in the field of probabilistic graphical modelling, DAGs are yet to be widely adopted in applied health research, where causal assumptions are frequently made for the purpose of evaluating health services initiatives. In this context, there is still limited practical guidance on how to construct and use DAGs. Some progress has recently been made in terms of building DAGs based on studies from the literature, but an area that has received less attention is how to create DAGs from information provided by domain experts, an approach of particular importance when there is limited published information about the intervention under study. This approach offers the opportunity for findings to be more robust and relevant to patients, carers and the public, and more likely to inform policy and clinical practice. This article draws lessons from a stakeholder workshop involving patients, health care professionals, researchers, commissioners and representatives from industry, whose objective was to draw DAGs for a complex intervention—online consultation, i.e. written exchange between the patient and health care professional using an online system—in the context of the English National Health Service. We provide some initial, practical guidance to those interested in engaging with domain experts to develop DAGs.

While machine learning (ML) methods have received a lot of attention in recent years, these methods are primarily for prediction. Empirical researchers conducting policy evaluations are, on the other hand, preoccupied with causal problems, trying to answer counterfactual questions: what would have happened in the absence of a policy? Because these counterfactuals can never be directly observed (described as the “fundamental problem of causal inference”) prediction tools from the ML literature cannot be readily used for causal inference. In the last decade, major innovations have taken place incorporating supervised ML tools into estimators for causal parameters such as the average treatment effect (ATE). This holds the promise of attenuating model misspecification issues, and increasing of transparency in model selection. One particularly mature strand of the literature include approaches that incorporate supervised ML approaches in the estimation of the ATE of a binary treatment, under the unconfoundedness and positivity assumptions (also known as exchangeability and overlap assumptions). This article begins by reviewing popular supervised machine learning algorithms, including trees-based methods and the lasso, as well as ensembles, with a focus on the Super Learner. Then, some specific uses of machine learning for treatment effect estimation are introduced and illustrated, namely (1) to create balance among treated and control groups, (2) to estimate so-called nuisance models (e.g., the propensity score, or conditional expectations of the outcome) in semi-parametric estimators that target causal parameters (e.g., targeted maximum likelihood estimation or the double ML estimator), and (3) the use of machine learning for variable selection in situations with a high number of covariates. Since there is no universal best estimator, whether parametric or data-adaptive, it is best practice to incorporate a semi-automated approach than can select the models best supported by the observed data, thus attenuating the reliance on subjective choices.