A decision kit
A free kit for one conversation: you, your cardiologist, and the published evidence on Eliquis (apixaban), turned into numbers for a patient like you. Nothing to buy, nothing to sign up for, and nothing in it that asks you to change anything on your own.
If you have atrial fibrillation, you have probably been prescribed Eliquis at one of two doses: the full dose of 5 mg twice a day, or the half dose of 2.5 mg. The label's rule for who gets the half dose looks precise, but no clinical trial has ever compared the two doses head to head, and in practice most people on the half dose do not actually meet the rule's criteria. Meanwhile the comparison that matters most to a patient, this drug against not taking it, is one the trials never ran directly.
The kit puts numbers on those questions for a particular person. At its center is a Bayesian model built from the published trials, which takes six facts a cardiologist reads off a chart, sex, age group, weight, a kidney-function value, the pattern of the fibrillation, and any history of stroke or mini-stroke, and returns that person's yearly chances of stroke and of major bleeding under three choices: full dose, half dose, or no anticoagulant. Every number carries a range, because the evidence itself has gaps, and the kit shows those gaps instead of hiding them.
One instruction stands above everything here: never stop a prescribed blood thinner, or change your dose, on your own. Going without protection is the one choice the evidence condemns without ambiguity. The kit exists to start a conversation with your cardiologist, not to replace one.
Start here, for you
Four pages in plain language. It shows what a year looks like among 1,000 people like one example patient under each choice, explains why the drug's added bleeding is mostly the survivable kind while the strokes it prevents are often not, and gives you three questions worth asking at your next appointment.
For the two of you
An Excel workbook with six dropdown menus and one results table. Your cardiologist sets the six values from your chart, which takes about a minute, and the table and chart then show your numbers, with their ranges, under full dose, half dose, and no anticoagulant. It is locked so that only the six menus can be changed: nobody can break it, nothing can be quietly altered, and every formula stays visible for a doctor who wants to check the work.
For your cardiologist
Written in the language of the clinical literature. It names every trial the model draws on, states every assumption and where judgment was required flags it as judgment, reports how the model was validated, and shows which pieces of evidence carry the conclusions. Doctors are rightly skeptical of printouts from the internet; this paper is built for that skepticism.
For your cardiologist, in a hurry
One page: what the model is, what goes in, what comes out, what it was checked against, and what it deliberately leaves out. Enough for a busy clinician to decide in two minutes whether the spreadsheet is worth the one minute it takes to run.
The full story
The paper the kit grew out of, for the curious reader of any background. Where the 2.5 mg rule came from, why the drug should be measured against not taking it, and how the model turns published trials into personal numbers, worked through step by step by counting.
Read the patient guide first; it is short and it is written for you. Then bring two things to your next appointment: the technical paper (printed or as a link) and the address of this page, so your cardiologist can open the spreadsheet. In the visit, your cardiologist sets the six menus from your chart, and the two of you are then looking at the same screen, talking about your numbers rather than about numbers in general.
The guide suggests three questions for that conversation. If you are on the half dose: do I actually meet the label's criteria, and if not, what is the reason for my reduced dose? If you are on the full dose and worried about bleeding: what would I give up by reducing, and does anything about me change that? And whatever your dose: is there anything about my situation this model does not capture? That last question is the most important one, because the answer is almost certainly yes, and it is where your cardiologist's judgment does the work no spreadsheet can.
It is a summary of published evidence, organized honestly. It is not medical advice, and it does not know you. The model deliberately leaves out things your cardiologist will weigh: other medicines that interact with Eliquis, your risk of falling, prior serious bleeds, liver disease, what happened the last time you were on a blood thinner. Its numbers are one-year averages over people who resemble you on six characteristics, and you are more than six characteristics. That is why every piece of the kit points to the same place: the conversation.