How these pages are written
When you’re living with AFib, it’s hard to know which claims to trust. These are the rules every research page follows, so you can judge them for yourself.
Which sources count
A source earns its place by answering a specific question. Randomized trials can test an intervention; observational studies can describe associations; device validations test particular measurements. A review inherits the limits of the studies it includes. Guidelines explain clinical context. Manufacturer testing and conference abstracts are labelled separately. Headline results are checked against the original paper, its abstract or official documentation; secondary summaries are used only to locate and understand sources.
Every source in the list at the end of a page says what kind of evidence it is. A finding from one small survey is never written as if it were settled.
How numbers are reported
- Each finding names the population, comparison, outcome, time period and study design.
- Developing AFib, acute episode associations, recurrence, burden and symptoms are different outcomes and are kept separate.
- Figures come from the original source. Confidence intervals and absolute results are included when they help interpret the finding.
- When a study found that two things go together, the page says they were linked. It says one thing causesanother only when trials have shown it.
- A percentage rise in risk is a relative figure. Where it matters, the page says so, because a 50% rise in a small risk is still a small risk. Absolute comparisons require results for the actual groups being compared, not an overall cohort average.
- Every number links to its source, so you can read the original.
What these pages won’t do
- Tell you to change a medication, a treatment or a habit. That’s a conversation for your care team.
- Diagnose, predict an episode, or turn a pattern in a personal record into proof of a cause.
- Treat symptom times, spot ECG recordings and true episode onset as interchangeable.
- Recommend products, carry ads or take sponsorship.
Who writes them, and how they’re checked
I’m Josh Adam, and I build Evenbeat. I have a background in microbiology and laboratory work, I’m not a doctor, and I live with AFib. These pages are researched and drafted with the help of AI tools. Each source has an evidence record with the result, population, key limitation, source location, access scope and date checked. That record distinguishes a full-text check from an abstract or documentation check; AI-assisted checking does not mean a clinician has reviewed the page.
Each page says whether a clinician has reviewed it, with a separate date if so. Pages without that review say so at the top. Bibliographic identifiers and visible correction or retraction notices are checked in the consulted records. An unresolved publication-status issue is recorded internally and that source is held out of published articles until resolved or replaced. A record check cannot guarantee that every notice has been found.
Dates and corrections
Each page separates publication, update, evidence-check and clinical-review dates. The review target is at least once a year or when an important new study appears. If you find an error, a broken link or a newer study, write to support@evenbeat.app. Corrections are made quickly and noted on the page with their date.