Which obesity treatments fit you?
Answer a few questions. COMPASS shows which treatments fit your situation, how much weight people lost with each one in research studies, and the main risks, so you can talk it through with your care team. It does not tell you what to do.
- Which options fit you, and why
- Weight lost in studies, in pounds for your weight
- The main risks, shown as "X in 100 people"
Nothing you enter is saved or sent. It stays in this browser tab. An optional survey at the end sends only its own answers, and the site counts, with no details about you, how often results are shown.
Your information
Step 1 of 4
Try an example
For demonstrations: fills in the form for a made-up person and shows the results. You can change the answers afterwards.
Your results
Your options at a glance
How much weight people lost
Drag to compare your goal with the study results below (the dashed line). It does not change which options fit you.
Your options
More options
Compare two options
Show the research behind these numbers
For clinicians: which studies
Key studies: randomized trials first, then large cohorts where fewer than two trials report a time (see "About the evidence and rules"). All studies: every value that passes the chart rules, including post hoc reports, other doses and diabetes-only trials. The switch changes every number on this page.
Side by side
Each study's result, by time
Your answers
Help improve COMPASS (optional, about 1 minute)
Answering is optional and anonymous. Only the answers in this box are sent. Nothing you entered above, such as your weight or health conditions, is included. The COMPASS project team uses the answers to improve COMPASS and may publish them, only as totals.
When to ask for a specialist referral
Ask your primary care clinician about seeing an obesity medicine or bariatric program if any of these apply.
- Your BMI is 35 or higher, or 30 or higher with type 2 diabetes. Guidelines recommend that people in these groups be told about surgery. ASMBS/IFSO 2022
- Your BMI is 30 to 34.9 and your weight-related conditions are hard to control, or other treatments have not worked well enough. Surgery may be considered. ASMBS/IFSO 2022
- You are 13 to 17 years old. Treatment for teenagers should be directed by a specialist team. ASMBS 2018
- You have had weight-loss surgery and are regaining weight, or have new symptoms such as repeated vomiting, severe reflux, or pain after eating.
- You have a safety flag in this tool (for example thyroid cancer history, pregnancy plans, or mental health symptoms that are not well managed right now), or you are unsure whether a medicine is safe with your other medicines.
- Severe obesity began in early childhood, or came with constant intense hunger. Rare genetic causes need a specialist. This tool does not cover them.
About the evidence and rules
How to read the numbers
- Each number is the average weight loss reported by one study, as a percent of starting weight. A range shows the lowest and highest of these. Nothing is averaged across studies, and it is not a prediction for you.
- The strongest studies are used first: randomized trials, then large long-term studies where trials are lacking. Two clinicians checked each number against the original paper.
- If you have type 2 diabetes, studies done only in people with diabetes are shown on their own line. People with diabetes often lose somewhat less weight.
- Studies done in Asian populations are used only if you tick Asian ancestry.
- If you are 13 to 17, the numbers come from studies of teenagers where they exist.
For clinicians: methods, evidence set and rules
Eligibility comes from FDA labels and the 2022 ASMBS/IFSO and 2018 ASMBS pediatric guidelines, and preferences re-sort options within what is eligible. Weight loss and risks come from key studies, each number traced to a quoted source, never pooled, modelled or predicted. COMPASS does not replace clinical judgment, and it does not assess cost, coverage or local program availability.
- Key studies: for medicines, the main report of a randomized trial at the dose approved for weight loss, in adults with obesity; for procedures and lifestyle programs, randomized trials and their long-term follow-up, plus large cohorts and registries. Randomized trials come first: where at least two report an option at a time, cohort studies are left out for that time; where fewer do, large cohorts fill the gap. A trial reported more than once at the same time point counts once. All other studies stay in the detailed tables. Studies limited to people with type 2 diabetes are never mixed into the main row for any option; users with diabetes see them in their own row. Users aged 13 to 17 see adolescent studies instead.
- Charts and summaries use only average percent total body weight loss at a clear time for the whole group studied, from the extracted text of each paper (a value may differ from the paper in format, for example a unit added). Other measures (kilograms, the share of people reaching a goal), subgroups, reviews that pool other studies, and people who had surgery or another treatment before are in the tables only. Each chart lists the options from most to least weight loss at that time (median of the values shown). The side-by-side table shows about 1 year and the longest follow-up. The study charts under it show every time window.
- Key-study rules: medicines need a primary RCT report (post hoc or secondary reports only from the 2-year window on), an FDA-approved obesity dose when the arm names a dose, and not a lead-in-responder population; adolescent studies are used only for users aged 13 to 17. Procedures and lifestyle: RCTs and their follow-up reports (for the bypass, standard RYGB arms only, not banded, distal or long-limb), prospective cohorts or registries of 100 or more (50 or more for adolescent studies), retrospective cohorts of 1,000 or more (lower cut-offs would add only small single-center endoscopic series; a higher prospective cut-off would remove all SADI-S evidence). One report per trial per time point, so a trial's 5- and 10-year reports both count. Trials first: in each time window, when 2 or more randomized trials report an option, its cohort and registry values are left out of that window in the key-study views; the tiles say how many. Trials limited to type 2 diabetes are marked separately: kept out of every main summary and shown in their own row for users with diabetes. The "All studies" switch shows everything that passes the chart rules, with no trials-first step.
- A supplement adds landmark results the extraction missed, for all seven review topics: found with Consensus, written from the PubMed abstract or PubMed Central full text, and checked by script word for word against those records. Examples: SLEEVEBYPASS, SM-BOSS at 10 years, STEP 1, OASIS 4, SURMOUNT-1 at 3 years, SURMOUNT-5, ATTAIN-1, LABS-2, PCORnet 2018, MERIT, the balloon pivotal trials. They pass through the same rules as the main table. Rows found wrong are left out. The head-to-head boxes and the after-stopping lines are checked against quoted source sentences when the data are built.
- Certainty labels come from a transparent rule set (study design, post hoc or secondary analysis, reported risk-of-bias language, directness, precision). They are not a formal GRADE assessment.
- Evidence was extracted with AI tools and checked by script against a quoted sentence from its source. Then two clinicians each checked every number in the summaries, charts, risk lists, head-to-head boxes and after-stopping lines against the original paper, without seeing each other's answers. Any number they did not both mark correct was decided by the project lead. Extraction can miss values, so "no studies report this" means none in this evidence set, not none in the world.
- Rows whose effect measure and outcome field disagreed were left out of this tool. Papers whose citation does not match their content are flagged and never charted.