The heterogeneity of treatment effects faced in real world study of syndromes in critical care, the limited RCT reproducibility over the past 3 decades, and the emergence of the ability to process massive data sets, has led to a move toward greater acceptance of observational trials.
Here is an ATS statement relating to this important development.
Of course, with loss of randomization, the limitations of causal inference based on obs trials are well known. A move toward measurement of dose response and timeseries of the pattern of response in relation to the time pattern of treatment will need to be incorporated. In otherwords, the DAG is expanded to engage relational time patterns (RTP). The relational/temporal patterns are the pivotal data of Obs trials and the analysis of these must be improved if obs trials are going to take a seat closer to RCT.