Log in with your email address username.


[Comment] Improving observational studies in the era of big data

Although randomised controlled trials (RCTs) predominate in the assessment of new medical technologies, large observational studies have grown popular in recent years, with widespread enthusiasm for big data approaches (eg, electronic health records). Observational studies provide advantages over RCTs, although limitations to these studies when using big data must be acknowledged and, when possible, addressed. Three common threats to valid inference from observational studies using big data include residual confounding, time-zero issues, and multiplicity.