By using this website, you agree to our use of cookies (needed to provide our website’s features) and our Terms of Use.
By: Andrew Forrest - January 2026
This guide explains the most common types of health studies, what they're suitable for, and how to assess the strength of the evidence, which is especially useful when you're reading studies on the health benefits of walking.
When you read claims such as 'walking lowers blood pressure', 'walking helps you sleep better', or 'walking reduces anxiety', it helps to know how those conclusions were reached. Different study designs answer different questions, and some are much better at showing cause and effect than others.
This page contains affiliate links. If you buy products or services via these links, we may earn a small commission at no cost to you. If you are thinking of buying, please use our links, as it helps support our website and YouTube channel.
Two studies can examine the same topic and reach different conclusions because of:
A major goal of good study design is to reduce confounding and bias, so we can trust that the results reflect reality.
For the question 'Does an intervention work? (e.g., Does starting a walking programme improve health outcomes?), evidence often becomes more reliable as you move up this ladder:
Important nuance: hierarchies are helpful, but you still need judgment, and a systematic review is only as good as the studies it includes.
What it is: A systematic, structured search and evaluation of all relevant studies on a question. A meta-analysis statistically combines the results across studies.
Why it's strong: It considers the totality of evidence (not just one study) and can explain why results differ.
What to watch for: Poor-quality included studies, missing unpublished studies, and mixing very different interventions/outcomes ('apples and oranges').
Walking example: A review pooling multiple randomised controlled trials of walking programmes to estimate the average change in blood pressure.
These are commonly considered the best 'top-of-the-ladder' evidence when they are well-conducted.
What it is: Participants are randomly allocated to an intervention (e.g., a walking programme) or a comparison (e.g., usual care, stretching, education, or another activity).
Why it's strong: Randomisation helps make groups similar on average, reducing confounding - so differences at the end are more likely to be due to the intervention.
What to watch for:
Walking example: Adults are randomised to a 12-week brisk-walking programme or a control group, and blood pressure is measured before and after.
Blinding means that someone doesn't know which group a participant is in to reduce bias.
Why it matters: If people know they're 'in the walking group', they might report feeling better (expectation effects), and assessors might (even unconsciously) measure differently.
Reality check for walking studies: It's often hard to blind participants to exercise. So strong walking trials often use:
Guidance on what RCT reports should include (including blinding and participant flow) is captured in CONSORT reporting guidance. CONSORT reporting guidance is a standardised checklist and flow diagram that helps researchers clearly and transparently report how a randomised controlled trial was designed, conducted, analysed, and interpreted.
What it is: An intervention is provided, but the assignment isn't random. Examples: before-and-after studies with a comparison group, natural experiments, and interrupted time series.
Why it's useful: Randomisation is sometimes impractical (e.g., a community walking trail built in one area).
What to watch for: Differences between groups at baseline, time trends, and other simultaneous changes.
Walking example: One workplace introduces walking breaks, while another does not.
What it is: Researchers follow a group over time and compare outcomes between people with different exposures (e.g., higher versus lower walking levels).
Why it's valuable: Great for long-term outcomes, rare exposures, and 'real-world' patterns - often with very large samples.
What to watch for: Confounding (walking may be linked to many other healthy behaviours), measurement error (self-reported activity can be inaccurate).
Walking example: Following 50,000 adults for 10 years to see whether those who walk more have lower rates of type 2 diabetes.
What it is: Start with an outcome (e.g., heart disease) and look backwards to compare prior exposure (walking) between cases (with disease) and controls (without).
Why it's useful: Efficient for rare outcomes or outcomes that take a long time to develop.
What to watch for: Recall bias (people may misremember past activity), selection of appropriate controls.
What it is: A 'snapshot' that measures exposure and outcome at the same time.
Why it's useful: Quick, inexpensive, and good for describing patterns and generating hypotheses.
What to watch for: You usually can't tell which came first (did less walking lead to worse health, or did worse health lead to less walking?).
Walking example: Surveying adults to see whether people who report more walking also report a better mood - useful, but not proof of causation.
These observational designs - cohort, case-control, and cross-sectional - are the main focus of STROBE reporting guidance. STROBE reporting guidance is a checklist that helps researchers transparently and comprehensively report observational studies (cohort, case-control, and cross-sectional designs).
What it is: A case report describing one patient or a case series describing a small group.
Why it's useful: Early signals, unusual effects, and new hypotheses.
What to watch for: No control group - cannot infer causation.
What it is: Interviews, focus groups, or observations exploring experiences, barriers, and motivations.
Why it matters: Doesn't answer 'does it work?' but answers:
This is often essential for designing realistic walking programmes.
Different questions call for different 'best' study types:
Regardless of the design, stronger evidence tends to have:
When evidence is summarised (especially in systematic reviews and guidelines), a common approach is to use a structured system called GRADE, which rates certainty (often: high, moderate, low, very low).
Certainty can be lowered if there's:
This provides a structured way to say not just 'what the studies found', but 'how sure we are'.
Research strongly suggests that walking supports health, but most studies focus on specific step counts and groups, so results may not apply equally to everyone.
Benefits observed in 'more-walking' groups may also reflect other healthy habits (such as diet, sleep, and stress management), making it hard to establish cause and effect. Despite these limitations of studies, the growing body of research consistently points in the same direction: walking is an important part of overall well-being.
Happy evidence-based walking. 😊
January 2026
