In asking the right questions, be cautious of confounding by indication

Excerpt

If you are admitted to an intensive care unit (ICU) today, your chances of survival are better than they have ever been. In the past three decades, there has been a consistent reduction in ICU mortality, irrespective of the reason for admission.1 For example, in 2000, a patient admitted to an ICU in Australia and New Zealand with severe sepsis had a 35% risk of mortality.2 This absolute risk of dying had dropped to 18% by 2012. It is hard to point to a single intervention that can account for this marked improvement in outcome. In fact, many of the interventions that were enthusiastically adopted during this time (e.g., intensive insulin therapy, activated protein C, early goal-directed therapy) have subsequently been shown to be ineffective3,4 or even harmful.5 What then is driving this dramatic decrease in mortality? Rather than a single intervention, these improved outcomes are likely the result of the small incremental benefits from multiple factors and improved processes of care.6 …