Abstract
We estimated the degree to which language used in the high-profile medical/public health/epidemiology literature implied causality using language linking exposures to outcomes and action recommendations; examined disconnects between language and recommendations; identified the most common linking phrases; and estimated how strongly linking phrases imply causality. We searched and screened for 1,170 articles from 18 high-profile journals (65 per journal) published from 2010-2019. Based on written framing and systematic guidance, three reviewers rated the degree of causality implied in abstracts and full text for exposure/outcome linking language and action recommendations. Reviewers rated the causal implication of exposure/outcome linking language as None (no causal implication) in 13.8%, Weak 34.2%, Moderate 33.2%, and Strong 18.7% of abstracts. The implied causality of action recommendations was higher than the implied causality of linking sentences for 44.5% or commensurate for 40.3% of articles. The most common linking word in abstracts was “associate” (45.7%). Reviewer’s ratings of linking word roots were highly heterogeneous; over half of reviewers rated “association” as having at least some causal implication. This research undercuts the assumption that avoiding “causal” words leads to clarity of interpretation in medical research.
Introduction
In observational health research, a long-standing "taboo" exists against using explicit causal language (e.g., "X causes Y") due to the inherent risks of confounding and bias in non-experimental designs. To maintain scientific caution, researchers frequently adopt associational euphemisms, such as stating that an exposure is "associated with" an outcome. However, this linguistic strategy creates a fundamental tension: if a study does not claim a causal relationship, the justification for clinical or policy action recommendations based on those findings becomes unclear. This study evaluates whether avoiding causal terminology actually improves clarity or if it merely allows for "Schrödinger’s causal inference," where findings are simultaneously treated as mere correlations and as the basis for practical intervention.
Methodology
The authors conducted a systematic evaluation of 1,170 articles (1,065 observational studies and 105 Randomized Controlled Trials for comparison) published between 2010 and 2019 across 18 high-profile journals, including The Lancet, JAMA, and NEJM.
A team of 48 reviewers, primarily experts in epidemiology and statistics, assessed each article. Each paper was reviewed by three unique randomly selected reviewers (two independent and one arbitrator). Reviewers used a standardized four-point scale to rate the "causal implication strength" of both the linking sentences (describing results) and the action recommendations (prescribing behavior).
Key Results
The "Association" Trap
The word "associate" (or its derivatives) was the most common linking term, appearing in 45.7% of abstracts. Despite its intended use as a non-causal descriptor, the study found significant linguistic heterogeneity. Over half of the reviewers rated the word "association" as having at least some causal implication, suggesting that the technical vocabulary of epidemiology is highly subjective and often interpreted causally by the research community.
Disconnect Between Language and Action
A primary finding was the disconnect between how results are described and how they are applied. In 44.5% of the articles, the action recommendations implied a stronger causal link than the sentences used to describe the primary results. Researchers often employed "cautious" language in their findings while offering "bold" clinical advice, effectively implying causality without meeting the rigorous methodological standards typically required for such claims.
Implied Causality in Observational Research
Even when avoiding explicit causal terms, only 13.8% of linking language in observational abstracts was rated as having "No" causal implication. The majority were rated as Weak (34.2%), Moderate (33.2%), or Strong (18.7%). Furthermore, while only 3.8% of studies utilized formal causal models like Directed Acyclic Graphs (DAGs), 80% included theoretical discussions of causal mechanisms, indicating that causality is almost always the underlying intent of the research.
Discussion and Conclusion
The results suggest that the current practice of "policing" causal language does not lead to greater clarity. Instead, it fosters ambiguity and allows for the presentation of causal conclusions under the guise of associational language. Interestingly, Randomized Controlled Trials (RCTs) used similar linking language to observational studies, suggesting a universal "scientific dialect" that obscures the actual strength of the underlying evidence.
To improve research communication, the authors recommend moving away from linguistic gymnastics. Instead of focusing on which words to avoid, researchers should clearly state their target estimand—explicitly defining whether they are seeking a causal effect or a simple prediction. Clarity in medical research is a product of methodological transparency and explicit intent, not the avoidance of specific vocabulary.
Summary updated