Launch offer: full deHype evidence reports are £4.99 for a limited time.

Check a headline

Evidence type

Observational studies

Observational studies look for associations in real-world populations but do not automatically prove that one factor causes another. Other explanations—including confounding and reverse causation—are important to consider when interpreting their results.

Observational studies are a common type of research used in health and social sciences. In these studies, researchers observe groups of people in real-world settings without assigning them to different interventions. The goal is often to find associations between exposures (such as dietary habits, physical activity, or medication use) and outcomes (like disease rates or health events).

What observational studies can tell us

Observational studies can reveal whether people who are exposed to a certain factor tend to have different outcomes compared to those who are not. For example, they can find that people who exercise more often have lower rates of certain illnesses than those who exercise less. This information can be valuable for generating hypotheses and understanding patterns in large populations.

Limitations of observational studies

  • Confounding: Other factors may explain the association seen in the study. For example, people who exercise more might also have healthier diets or higher incomes, which could be the real reason for their better health.
  • Reverse causation: Sometimes it is unclear which came first—the exposure or the outcome. For example, people might start exercising less because they are already becoming unwell, not the other way around.
  • Selection bias: The people studied may not be representative of the entire population, affecting how generalisable the findings are.
  • Measurement error: Factors like self-reported habits or incomplete data can influence results.

Why observational studies can't prove causation alone

Even if a strong association is found, it does not automatically show that the exposure causes the outcome. Proving cause and effect usually requires additional evidence, such as from randomised controlled trials or other study designs that minimize confounding.

When observational studies are valuable

  • They are often the only practical way to study factors that cannot be ethically or feasibly randomised (like air pollution or long-term lifestyle habits).
  • They can reveal possible risks or benefits worth testing more rigorously in future studies.
  • They provide important real-world data about health trends and associations at the population level.

Summary

Observational studies are useful for detecting associations but have important limitations. The findings can guide future research but should not be taken as definitive evidence of causation without considering confounding and other biases.

Reports

Related deHype reports

Specific public claims connected to this topic.

Report · Grade D Evidence check: Poor sleep linked to rising cancer risk in under-50s The Guardian article reports on two large US studies, presented at a major cancer conference, that found an association between poor sleep and increased rates of cancer in under-50s. While people diagnosed with insomnia appeared to have up to three times higher risk of some cancers within five years, experts stress that these are observational findings and do not prove that poor sleep causes cancer. Potential confounding factors (such as drinking, smoking, and obesity) and the possibility of reverse causality (cancer affecting sleep before diagnosis) are noted. Further research is needed to clarify the relationship and potential biological mechanisms. Report · Grade C Evidence check: Blood and spinal fluid proteins reveal distinct fingerprints of four brain diseases A large-scale, multi-institutional study analyzed nearly 7,000 proteins from blood and spinal fluid samples of close to 6,000 participants, identifying disease-specific and shared molecular signatures for four major neurodegenerative diseases. These protein patterns enabled computational models to differentiate between diseases and controls with high accuracy and may guide the development of future diagnostic tests and therapeutic targets. However, clinical application and real-world benefit remain unproven. Report · Grade C Evidence check: Cycling Linked to Lower Risk of Dementia, Study of Half a Million Finds A large observational UK cohort study found that individuals who primarily cycled for travel had a lower risk of all-cause dementia, Alzheimer's disease, young-onset and late-onset dementia compared to non-active travelers. The study also linked cycling to greater regional brain volumes relevant to memory, but noted important confounders and did not demonstrate causality. Report · Grade C Evidence check: Early-Onset Colorectal Cancer Risk Factors Identified The article describes two large observational studies, each analyzing risk factors for early-onset colorectal cancer (CRC) and colorectal adenomas (<50 years). The first, a case-control analysis using a large health database, found that younger adults with early-onset CRC were more likely to have inflammatory bowel disease, family history of CRC, obesity/severe obesity, metabolic syndrome, diabetes, hepatic steatosis, and certain lab abnormalities compared with those with late-onset CRC and with healthy controls. The second, a nested case-control study in the Kaiser Permanente system, found a statistically significant association between extended oral antibiotic use and the risk of developing early-onset colorectal adenomas, independent of class and duration, with risk increasing with greater use. Both studies are observational, cannot prove causation, and were presented in conference abstract form, not yet in peer-reviewed full publication. Commentary from clinical experts in the article emphasizes the hypothesis-generating nature of these findings and cautions against changes in clinical practice or screening recommendations until further research is completed.

Helpful evidence guides

How deHype grades claims Learn how deHype evaluates health and science claims with a transparent grading system based on strength, relevance, and quality of evidence. How to read a health headline A practical guide to spotting overstatement and understanding the evidence behind health and science headlines.

Related evidence pages

Animal studies Animal studies can provide valuable insights into biological processes and potential effects of interventions. However, results from animal experiments may not translate to humans. This page explains why animal research can inform but not confirm health claims about people. Cell studies How in-vitro and cell-line findings should be interpreted in health headlines. Cell studies investigate the effects of substances or genetic changes in isolated cells outside the human body. While these studies can show how something may work at a basic biological level, their results do not directly apply to human health without further research. Guidelines Guidelines translate research evidence into recommendations for clinical practice, but their strength can vary depending on the available evidence, expert interpretation, and other practical factors. Randomised trials Randomised trials are a key method for testing the effects of medical and health interventions. By randomly assigning participants, these studies aim to balance differences between groups and reduce bias, making their findings more reliable. However, they are not free from limitations, and careful interpretation is still needed. Systematic reviews Systematic reviews summarise multiple studies using transparent, pre-specified methods. Their strength as evidence depends not only on their process but also on the quality and relevance of the included studies.
Seen a headline in this area? Paste the link and get a deHype evidence check.
Check a headline