The headline number from the Phase 3 TRIUMPH-4 trial is the weight loss: up to 28.7% mean body weight reduction at 68 weeks on retatrutide 12 mg, against 2.1% on placebo. That figure is the highest reported for any obesity drug at the Phase 3 stage. The number that's quieter — and arguably more useful for anyone weighing the decision to start one of these drugs — is the knee pain reduction: a 4.5-point drop on the WOMAC pain subscale, which works out to roughly a 75.8% reduction relative to baseline. About 14.1% of the 9 mg arm and 12% of the 12 mg arm were free of knee pain at 68 weeks, compared with 4.2% on placebo.
If you're a man in your 30s or 40s with knees that have been complaining for a few years and a clinician suggesting a GLP-1 or triple-agonist class drug for weight management, that secondary endpoint is the one your day-to-day life will probably notice first. Bigger lifts feel different at week 12 of a cut than week 1. Stairs feel different at month 4. The TRIUMPH-4 data says, on average, those subjective signals match what the formal pain instruments measure. Whether they match for you — your knees, your training history, your particular mix of mechanical and inflammatory pain drivers — is a question only your data can answer.
This post is about that personal data. It's not a recommendation about whether to use these drugs. It's about what to log if you do.
Retatrutide is investigational and not FDA-approved. Semaglutide and tirzepatide are FDA-approved but knee osteoarthritis is not their on-label indication. Decisions about whether, when, and how to start any of these therapies belong with you and the prescriber who knows your full medical history. The framing of this post is educational, not prescriptive.
What TRIUMPH-4 Actually Found
The trial enrolled 445 adults with obesity or overweight (without diabetes) and clinically diagnosed knee osteoarthritis, randomized 1:1:1 to retatrutide 9 mg, 12 mg, or placebo over 68 weeks. The primary endpoint was percent body weight change. The secondary endpoint was knee pain via WOMAC. Both endpoints moved in the same direction at meaningful magnitudes.
The mechanism question is open. The trial wasn't designed to disentangle "is the pain reduction caused by weight loss alone, or is there a separate anti-inflammatory effect of the drug?" Knee pain in people with obesity is mechanically and inflammatorily linked to body weight, and the published OA guidelines have long included substantial weight loss as a non-pharmacological intervention. The 75.8% pain reduction is consistent with — and proportional to — the magnitude of weight loss observed. Whether retatrutide produces additional joint-specific anti-inflammatory effects beyond what weight loss alone explains is a question for trials specifically designed to answer it.
For the practical question — if I start one of these drugs, will my knees feel better? — the mechanism question matters less than the dose-response relationship. If the pain reduction tracks weight loss for you, you have a feedback loop. The body weight number is the leading indicator; the pain number is the lagging one. Watching both gives you something more useful than either alone.
Why You Want a Pain Baseline Before You Start
The most common mistake people make when they start a high-impact intervention is failing to capture a clean baseline. The week before initiation is the highest-information week in the entire experiment, because every measurement after that is being interpreted relative to it. If your starting WOMAC is a vague memory of "my knees have been bad for years," the question "did this work?" is going to come down to vibes.
A useful baseline for joint-pain tracking is small enough that nobody skips it:
- WOMAC pain subscale, taken once. The five-question pain-only section of the WOMAC index, scored 0–4 per item (none, mild, moderate, severe, extreme), totaled to a number between 0 and 20. The instrument is free, validated, and self-administered.
- A 10-cm visual analog scale (VAS) for current knee pain. Not a great research instrument on its own, but it captures momentary state in a way the WOMAC doesn't.
- Functional check. A timed sit-to-stand (5 reps, recorded in seconds) and the highest squat depth that doesn't produce pain. Both are pre-rehab standards and both are quick.
- Steps per day for the 7 days before you start, pulled from your wearable. Not because steps cure knee pain, but because changes in step count after starting can confound the pain signal. (People who feel better walk more, which can produce a transient pain bump that isn't a setback.)
That's it. The week-before baseline is a 30-minute exercise, and it's the difference between knowing whether something worked and guessing.
What to Track Each Week During Titration
GLP-1 and triple-agonist class drugs typically follow a multi-month dose-escalation protocol. Side effects (nausea, fatigue, low energy) tend to concentrate during escalation; weight loss is gradual; pain change is gradual on top of the weight change. The right tracking cadence during this window is weekly, not daily — daily noise would drown the signal.
Once a week, same day, same conditions:
- WOMAC pain (5 questions, 2 minutes).
- Single VAS pain rating.
- Body weight, morning, after first bathroom visit, before food/drink.
- Waist circumference, same time as weight, no clothes interfering.
- Average daily steps for the prior 7 days (your wearable already has this).
- Resting heart rate average for the prior 7 days. Triple agonists in particular show modest dose-dependent RHR increases; this is worth watching.
Once a month, separately:
- Repeat the timed sit-to-stand.
- Re-test your highest pain-free squat depth.
- Take a single photograph of yourself in consistent lighting and a consistent pose. Useful for reading body composition shifts that the scale doesn't capture.
The week-to-week WOMAC trend is what you're watching. A meaningful clinical improvement is generally considered around a 2-point drop on the pain subscale. The TRIUMPH-4 average of 4.5 points was substantial — but averages are averages. Some people will improve more, some less. The point of tracking your own number is to find out which group you're in instead of arguing about it.
A practical heads-up about WOMAC. The full WOMAC has three subscales: pain (5 questions), stiffness (2 questions), and function (17 questions). The pain subscale alone is enough for a tracking journal. If you want a fuller picture, do the function subscale quarterly — it's more sensitive to slow recovery from chronic stiffness.
Confounders You Want to Log So You Can Rule Them Out Later
Six months into a weight-loss protocol with a complicated drug, you will inevitably ask the question: "Was it the drug? Was it the weight loss? Was it the new shoes? Was it the fact that I started swimming because cardio felt different?" The honest answer is usually some of all of the above. Logging the obvious confounders weekly is what lets you separate them out later.
Things to log alongside the pain numbers:
- Training changes. New activity, new shoes, change in resistance program, change in cardio modality. A simple line per week: "Added 2x/week pool sessions" is enough.
- Other medications and supplements. Especially anything you started or stopped during the protocol. NSAIDs are the obvious one — if your NSAID use changes, your reported pain will change for reasons that have nothing to do with the GLP-1.
- Sleep duration and quality. Sleep deficiency drives pain perception. A 1–5 daily sleep score plus your wearable's number is sufficient.
- Stress and mood. A 1–5 daily score. Stress and pain modulate each other through shared central pathways. You don't need a clinical instrument; you need consistency.
- Body weight method. Same scale, same day, same time. Switching from a home scale to a gym scale partway through is a confound.
The discipline isn't strict. A one-line weekly note in whatever app or notebook you use is plenty. The point is to have something to refer back to when you're trying to read a curve six months later.
What Counts as a Signal Worth Acting On
If you and your prescriber agreed on a 12-month protocol, the question of whether to keep going gets revisited along the way. Three patterns are worth flagging in advance:
The expected pattern. Weight comes down on schedule. WOMAC pain comes down a couple weeks behind it, lagging the weight curve. Resting heart rate either stays flat or creeps up modestly during titration and then settles. Functional measures (sit-to-stand, squat depth) improve. This is what TRIUMPH-4 average data describes; it's the trajectory most people in the trial saw.
The split signal. Weight comes down but pain doesn't move, or pain reduces but weight is stable. This is the most clinically interesting pattern because it suggests the mechanisms are partially independent in your case. A flat-pain-with-weight-loss trajectory at month 4–5 is worth a conversation with your prescriber and possibly an orthopedic referral — significant weight loss without the expected pain response can indicate a structural issue that won't resolve with conservative measures alone.
The reversal. Pain gets worse as weight comes down. This pattern is unusual but real, and it's almost always explained by the activity rebound — people feel better, walk and train more, accumulate volume their tissues haven't adapted to, and a new pain pattern emerges. If your weekly steps jumped 40% in the same window your WOMAC ticked up, the activity rebound is the first hypothesis. The fix is usually load management, not drug discontinuation.
The point of keeping the weekly log is that none of these patterns is obvious in real time. They're obvious in retrospect, looking at four months of numbers next to each other. The log is what makes retrospection possible.
How This Fits With Other Tracking You're Probably Already Doing
A lot of the inputs above are things you're already pulling from a wearable: weight, steps, RHR, sleep. The work the journal does is pulling those streams into one place alongside the pain instruments and the qualitative notes — which is the kind of thing wearable apps don't do well by default.
The Prova framing is a structured experiment around a specific question: did starting this drug improve my knees in a way that lasted? That's a four-data-stream question (drug, weight, pain, function) plus the confounder log, run for 6–12 months. The data needs to live somewhere you'll actually look at it later. A spreadsheet works. A dedicated tracking app works better if it lets you correlate the streams without copy-paste.
Be the first to try Prova
We're building an app to track whether GLP-1 joint pain tracking actually works. Join the waitlist.
What This Post Doesn't Cover
A few related questions this post deliberately doesn't try to settle:
- Should you start one of these drugs at all? Different question. It depends on individual circumstances no blog post can address — your weight, comorbidities, knee imaging, training history, and what you've already tried. That conversation belongs with your prescriber.
- Which drug? Semaglutide, tirzepatide, and retatrutide differ on efficacy magnitude, side-effect profile, dosing, and approval status. Retatrutide is investigational and only available through clinical trial enrollment as of May 2026. The framework above applies regardless of which agent your prescriber chooses.
- What if you can't get the drug? The non-pharmacological levers for knee OA in the setting of overweight — sustained moderate weight loss, resistance training, low-impact cardio, and structured rehab — also produce WOMAC improvements, just on a slower timeline. The same tracking framework applies.
Bottom Line
TRIUMPH-4 reported a substantial reduction in WOMAC knee pain alongside the weight loss everyone is talking about. The pain finding is consistent with what large weight loss has always been shown to do for knee pain in people with obesity, and the magnitude was large enough to matter clinically. Whether the same magnitude shows up in your knees depends on factors no average can predict — but those factors are mostly trackable.
A clean week-before baseline, a weekly WOMAC plus weight plus a few wearable metrics, and a confounder log is enough infrastructure to know whether what you're paying for is doing what the trial said it would.
The drug does one job. The data is the only thing that tells you whether the job got done.
Be the first to try Prova
We're building an app to track whether WOMAC pain tracking experiment actually works. Join the waitlist.