1. People do get better, by many different routes
Every account here describes some improvement, including people who had been ill for decades. They didn’t all do the same thing. If the first thing you try doesn’t work, that is the ordinary shape of these stories, not a verdict.
2. These are published stories, not trial results
Nobody counted the people who tried the same thing and stayed ill. There is no control group, and the illness sometimes improves on its own. “68% recovered” means “68% of the published accounts describe a recovery”, not “it works for 68% of people”.
3. Who publishes a story shapes what it says
Pages that sell a program, and advocacy sites that exist to share recoveries, publish the cases that worked. Forums and personal blogs have no editor, so the people who stayed ill post there too. Every treatment page shows outcomes by publisher so you can compare. On the explorer you can hide accounts from pages that sell the treatment.
4. Look at whether it lasted
Some treatments produce dramatic responses that fade when the treatment stops. Each treatment shows what share of accounts describe the benefit fading or relapsing, and what share needed to keep taking it. These are read from the text, so if a story simply stops without a follow-up it counts as “not faded”. Treat the faded figures as a minimum.
5. Every outcome is one reader’s judgement of the person’s own words
Severity, illness length and how complete the recovery was are as the person described them, and were coded by a reader. Spot checks found this is the least reliable part of the data: perhaps one account in seven would be graded differently on a second reading. The links and treatments are more reliable. Always read the original.
Where the stories come from
1,862 accounts from 134 websites: patient forums (mainly Phoenix Rising and Health Rising), personal blogs, recovery sites, news and podcasts. Accounts were found by reading forum sections and site indexes from start to end, not by searching for success stories. Reddit is not yet included.
Each source page has been downloaded again and archived. 10 accounts point to pages that are no longer online; for those, the site links to a copy on the Internet Archive.
Symptoms and side effects are found by scanning each summary for keywords, so those lists are incomplete. A story only appears under a symptom if its summary mentions it.