Causal inference methods for combining randomized trials and observational studies: a review

Bénédicte Colnet (1, 2) , Imke Mayer (3, 4) , Guanhua Chen (5) , Awa Dieng (6) , Ruohong Li (7, 8) , Gaël Varoquaux (1, 2) , Jean-Philippe Vert (9, 6) , Julie Josse (10, 11) , Shu Yang (12)

1 PARIETAL - Modelling brain structure, function and variability based on high-field MRI data
2 SODA - Méthodes computationnelles et mathématiques pour comprendre la société et la santé à partir de données
3 EHESS - École des hautes études en sciences sociales
4 Charité - UniversitätsMedizin = Charité - University Hospital [Berlin]
5 University of Wisconsin-Madison
6 Owkin France
7 Indiana University School of Medicine
8 Google Brain [Toronto]
9 Google Brain, Paris
10 AIDMP - Aide à la Décision pour une Médecine Personnalisé - Laboratoire de Biostatistique, Epidémiologie et Recherche Clinique - EA 2415
11 PREMEDICAL - Médecine de précision par intégration de données et inférence causale
12 Department of Statistics and Operations Research, University of North Carolina

With increasing data availability, causal effects can be evaluated across different data sets, both randomized controlled trials (RCTs) and observational studies. RCTs isolate the effect of the treatment from that of unwanted (confounding) co-occurring effects but they may suffer from un- representativeness, and thus lack external validity. On the other hand, large observational samples are often more representative of the target population but can conflate confounding effects with the treatment of interest. In this paper, we review the growing literature on methods for causal inference on combined RCTs and observational studies, striving for the best of both worlds. We first discuss identification and estimation methods that improve generalizability of RCTs using the representativeness of observational data. Classical estimators include weighting, difference between conditional outcome models, and doubly robust estimators. We then discuss methods that combine RCTs and observational data to either ensure uncounfoundedness of the observational analysis or to improve (conditional) average treatment effect estimation. We also connect and contrast works developed in both the potential outcomes literature and the structural causal model literature. Finally, we compare the main methods using a simulation study and real world data to analyze the effect of tranexamic acid on the mortality rate in major trauma patients. A review of available codes and new implementations is also provided.

Mots clés: causal effect generalization; transportability; external validity; data integration; heterogeneous data; double robustness

Domaines: Méthodologie [stat.ME]

Bénédicte Colnet, Imke Mayer, Guanhua Chen, Awa Dieng, Ruohong Li, et al.. Causal inference methods for combining randomized trials and observational studies: a review. Statistical Science, In press. ⟨hal-03008276v2⟩