Observational Studies in Health: Epidemiological Designs, Analytical Potential, and Limits in Causal Inference

Introduction: the choice of methodological design in health research constitutes an epistemological decision that conditions the validity, interpretability, and social utility of the evidence produced. Despite the widespread use of observational studies in epidemiology and public health, these designs are often treated in the literature in a fragmented manner and hierarchized uncritically, which may foster inappropriate causal interpretations and obscure their analytical limitations and potential. Objective: to analyze the observation study designs used in health science. Methods: this is a conceptual and interpretative literature review. The analysis was guided by classical and contemporary epidemiological frameworks, with emphasis on the concepts of temporality, validity, bias, and confounding. The Potential Outcomes Framework was adopted as a theoretical–interpretative reference to discuss the structural limits of causal inference in observational studies, without being applied as an analytical method. Results: the literature review indicates that classical observational designs do not constitute an absolute methodological hierarchy, but are suited to different research questions, depending on the time frame, population definition, and epidemiological measure adopted. Cohort studies preserve temporality and allow incidence to be estimated; case-control studies are more efficient for rare outcomes; and cross-sectional studies are used for prevalence estimates. All designs have structural limitations related to bias, confounding, and causal inference. Conclusion: observational studies are legitimate and indispensable methodological choices in health research, especially in contexts where controlled experimentation is ethically unfeasible or conceptually inappropriate. This study shows that the quality of observational inferences depends less on statistical sophistication alone and more on the consistency between the research question, the empirical context, the design adopted, and its theoretical assumptions. It is essential to recognize the structural limits of causal inference for the critical and responsible use of this evidence in public health policies.