Reading Clinical Studies: An Evidence Guide for Practitioners
Why Reading Skills Matter
Vendor material, social content and real evidence flow through the same channels (see information sources). The ability to read clinical studies is the divide between “I heard” and “I know”.
The Five-Minute Checklist
- Design: RCT > controlled study > case series > case report; prospective beats retrospective.
- Sample size and follow-up: n<20 is a lead, not evidence; follow-up shorter than the effect timeline (collagen studies under 3 months) gets discounted.
- Controls: uncontrolled “90% response” cannot attribute; split-face self-control is a common valid design in device studies.
- Endpoints: investigator-rated, subject-rated and instrument-measured are three different strengths — note which is used.
- Conflicts and venue: industry funding isn’t disqualifying but deserves a discount; peer-reviewed journals outrank abstracts and white papers.
Common Statistical Traps
- Relative games: “50% improvement” may be a 2-to-3 on a scale — check absolutes and baselines.
- Selection bias: responder-only photos; ask about dropout and intention-to-treat.
- Surrogate endpoints: dermal density up ≠ visible improvement — keep the two chains separate.
FAQ
Can I read studies without statistics training?
Yes — most judgment is design logic, not calculation; consult a research-literate peer when numbers (sample sizes) need evaluating.
Where to find device studies?
PubMed and CNKI are the main libraries; combine “mechanism + indication + RCT/systematic review” in searches, and read systematic reviews first.