Causal Language in Observational Orthopaedic Research.

Overview

With the increasing availability of large clinical registries and administrative data sets, observational (i.e., nonexperimental) orthopaedic research is being performed with increased frequency. While this research substantially advances our field, there are fundamental limitations to what can be determined through a single observational study. Avoiding overstatements and misstatements is important for the sake of accuracy, particularly for ensuring that clinical care is not inadvertently swayed by how an observational study is written up and described. We have noticed that causal language is frequently misused in observational orthopaedic research-that is, language that says or implies that 1 variable definitively causes another, despite the fact that causation can generally only be determined with randomization. In this data-backed commentary, we examine the prevalence of causal language in a random sample of 400 observational orthopaedic studies; we found that causal language was misused in 60% of them. We discuss the implications of these results and how to report observational findings more accurately: the word "association" (and its derivatives) can almost always replace or reframe a causal phrase.