New here? Start with how to use it, and why nobody can see your list.
Every grade answers one narrow question: how strong is the public evidence and safety record behind this drug, and how clean is its money trail and court record? It is not a verdict on whether a drug is right for you. A drug can be an F here and still be the right choice for a specific patient; a drug can be an A and still be wrong for you.
| Part | Weight | What moves it |
|---|---|---|
| Safety | 35% | FDA boxed warning, REMS safety program, controlled-substance status, how often deaths and serious outcomes appear in adverse-event reports compared with the FDA-wide average, off-label use dominating reports, Class I recalls |
| Evidence | 25% | How many randomized trials and systematic reviews exist, whether independent Cochrane reviews exist, whether completed trials actually posted their results, how many trials were abandoned, and whether "drug ineffective" is the top thing patients report |
| Regulatory | 15% | Recalls since 2016 (discounted for old generics with dozens of makers), REMS, how often the label has had to change recently, approval within the last five years |
| Money & lawsuits | 15% | Dollars drug companies reported paying US physicians and hospitals tied to this drug last year; whether it is named in pending federal multidistrict litigation |
| Options | 10% | Whether a generic exists and whether same-class alternatives exist, because options are leverage for the patient |
Letters: A is 87.5 and up, B 75, C 62.5, D 50, F below 50. In words: A means nothing serious on record; B, only the minor noise every long-used drug accumulates; C, one real concern; D, a boxed warning plus more; F, several serious problems at once. Every deduction is printed on the drug's page so you can see exactly why, and judge for yourself whether each one applies to you.
All of it is public and free: openFDA (labels, adverse events, recalls, approval history), the National Library of Medicine (RxNorm, RxClass, PubMed), ClinicalTrials.gov, CMS Open Payments, CMS Medicare Part D spending, and the Judicial Panel on Multidistrict Litigation. Pages are rebuilt automatically; the build date is in the footer.
497 drugs and fixed combinations: the most prescribed in Medicare Part D plus the all-ages top 300 from the national Medical Expenditure Panel Survey, so drugs mostly taken by younger people (birth control, ADHD, acne) are included too. Measured against the national top-300 list, the graded drugs account for about 97% of those prescriptions at every age, and about 86% of all Medicare prescriptions filled. The remaining gap is the long tail of rarely prescribed drugs. If a drug you take is not here, search for it anyway: the page tells you how to request it.
How well the drug works for your condition. The honest measure is the "number needed to treat": how many people have to take the drug for one of them to get a benefit they would not have had otherwise. For some drugs that number is 3; for others it is 100 or more. It depends on the condition, so ask your prescriber for it. Also missing: state-court lawsuits, foreign regulator actions, interactions with your other medications, and anything that is not in a public record. Adverse-event reports are voluntary and unverified; a report does not prove the drug caused the event.
Every grade on this site is an opinion, produced by the published method above from public records: FDA, National Library of Medicine, CMS, federal court dockets, and the named independent reviewers. Each page links to the records it was built from and states the date it was built. Facts are reported as the record states them: a settlement resolves allegations and is not a finding of wrongdoing; a pending lawsuit is not a verdict; an adverse-event report is not proof the drug caused anything. The site has no relationship with any drug maker, takes no advertising, and sells nothing. Pages are rebuilt from the primary sources on a schedule, and errors are corrected as they are found. Nothing here is medical advice.