Know what the evidence can support.
A useful reference shows the question, the comparison, the uncertainty and the source together.
The evidence pyramid
Design is a starting point, not a quality score. A review can be less reliable than the trials it includes.
The pyramid sorts study types
The levels help you navigate forms of research. They are not formal Oxford evidence levels or GRADE ratings. Risk of bias, precision, consistency and relevance still need assessment.
Oxford CEBM ↗Counts describe reports, not independent studies.
Five reading rules
Improvement does not always mean treatment benefit
Symptoms may change with time, rehabilitation, other care or expectations. Nonrandomized studies can also compare people with different prognoses. Their observations are useful, but causal claims need a suitable comparison.
Every graphic needs a source
Charts use named outcomes and timepoints. Calculated values are labelled, conflicting figures stay visible, and missing values are never replaced with zero. A study profile names the actual equipment only when verified.
Material consulted
Only analyses with verified full-text reading appear in this edition. Records based only on abstracts or selected sections are withheld. Each study page is then reviewed clinically by the editor, Dr. Marcus Yu Bin Pai.
What belongs in the library
Completed randomized trials, relevant nonrandomized human studies and reviews are included. Trial protocols, notice-only records and unrelated studies are outside the core count. Companion reports and overlapping reviews are not independent confirmations.
Units are part of the result
Pulses count applications; Hz describes pulses per second; mJ/mm² describes energy flux density; bar is a pressure setting. Device construction, applicator and measurement conditions matter. There is no universal bar-to-mJ/mm² conversion.