Real World Evidence vs Clinical Trial Results GLP-1: What the 2026 Data Shows

Quick look: trial vs. real-world weight loss

Bottom line: The drug mechanism doesn't change when you leave a trial — the treatment path around it does. If you stay on the drug for 12+ months, reach the maintenance dose, and have someone supporting the work, your number lives much closer to the trial range than to the cohort-wide average.

→ Use the Translation Score below to estimate which evidence column applies to your treatment path

What “real-world evidence” actually means for GLP-1s

Here's the cleanest way to think about it.

A clinical trial is a science experiment. Researchers pick the patients, set the dose schedule, supply the drug, and follow up on a fixed cadence. The point is to find out what the drug can do when conditions are tightly controlled.

Real-world evidence measures what actually happens after the drug leaves the lab. It pulls from sources like:

• Electronic health records (EHR): charts from your doctor's office.
• Claims data: what your insurance paid for and when.
• Pharmacy refill data: when you actually picked the drug up.
• Patient registries: structured databases that track patients over time.
• App-based programs: weight and adherence logs from digital health platforms.

Each source has blind spots. EHR data misses care outside that health system. Claims data shows fills, not whether you actually took the dose. Patient self-reports can be off. But put together, real-world evidence answers questions trials can't — like “do people actually stay on this drug?” and “how do results change when the prescription comes with a coaching program?”

The FDA takes this seriously. The agency uses real-world evidence to monitor drug safety after approval and, when the data is solid, to support new regulatory decisions. [22] Real-world evidence isn't a lesser cousin of trial data. It just answers a different question.

Why both matter: Trials estimate efficacy — what the drug can do. Real-world evidence estimates effectiveness — what the drug does do under everyday conditions. You need both to know what to expect.

Real world evidence vs clinical trial results GLP-1: what do the numbers show?

Master comparison table: trial vs. real-world weight loss

Every major pivotal trial and real-world dataset side-by-side, with source type shown on every row so you can weigh the evidence yourself.

Two real-world studies appear in our verified evidence list but not in the table because they reported relative or qualitative comparisons rather than table-grade percentages: Tchang et al. (Obesity 2026) [12] found semaglutide weight-loss outcomes among persistent telehealth users at 68 weeks comparable to clinical trial completers, and an Optum Market Clarity tirzepatide analysis (DOM 2025) [14] found 6-month weight reduction consistent with SURMOUNT-1 among persistent users.

The four reasons real-world numbers come in lower

Reason 1 — More than half of patients stop within a year

This is the single biggest driver. In the Truveta study of 125,474 US adults, 53.6% discontinued their GLP-1 within 12 months and 72.2% within 24 months. [9] By contrast, STEP-1 had a 1-year discontinuation rate of 17.1%, and SURMOUNT-1 was between 14.2% and 16.4%. [1][2]

When treatment stops, the trial-level weight-loss path usually stops too — unless something else replaces it, like restarting, switching, or structured lifestyle support.

Reason 2 — Most patients never reach the trial dose

The Cleveland Clinic study found that more than 80% of patients were on lower maintenance dosages — defined as ≤1 mg for semaglutide and ≤7.5 mg for tirzepatide. [7] The trial averages everyone cites were measured at the higher target doses (2.4 mg semaglutide; 5 mg, 10 mg, or 15 mg tirzepatide). [1][2]

Even in structured programs where cost wasn't a barrier, the climb to top dose is steep. In Samuels' academic obesity clinic, only 23% of semaglutide users reached 2.4 mg and only 28% of tirzepatide users reached 15 mg, despite a “no cost to patient” structure. [11]

Dose matters a lot. SURMOUNT-1 showed the dose-response clearly: −15.0% at 5 mg, −19.5% at 10 mg, −20.9% at 15 mg. [2] If treatment plateaus at a lower dose, the result usually does too.

Reason 3 — Trial support is structured; real prescribing usually isn't

The STEP and SURMOUNT trials required participants to receive structured lifestyle counseling, work toward a 500-calorie daily deficit, and complete at least 150 minutes of physical activity per week. [1][2][13] That counseling was free, scheduled, and continuous.

The real-world equivalent for most patients is the prescription, a short office visit, and “see you in three months.”

This is the gap that programs like Calibrate, Omada, and well-run telehealth services are starting to close. Calibrate's 2026 Results Report showed members lost 16% in year 1, 18% in year 2, 20% in year 3, and 21% in year 4. [17] That's trial-level weight loss sustained past trial-length follow-up. The drug didn't change. The support around the drug did. (Note: company-reported program data, not a randomized study.)

Reason 4 — Trial patients are pre-screened for adherence

Nobody likes to talk about this one, but it's real and the trials are open about it. To enroll in STEP-1 or SURMOUNT-1, you had to be willing to attend weekly or near-weekly visits and follow protocol. People who looked unlikely to do that didn't make it into the analyzed cohort.

That's not a flaw — it's how trials estimate what's possible under structured conditions. It just means the trial population was, on average, more committed than the population walking into a primary care visit.

Putting it together: Stop early + lower dose + no support + average commitment = the 8.7% routine-care average. Persist + target dose + structured support + high commitment = the 14–18% you see in trials. The drug mechanism behaves the same in both scenarios — the treatment path is what changes.

If you're looking for a provider that actively helps with dose escalation and prior authorization support, see our guide to best GLP-1 providers that help with prior authorization.

Wegovy vs Zepbound: how the gap compares by drug

What this tells you: Tirzepatide carries more weight-loss horsepower than semaglutide in both clean and messy data. The mechanism makes sense — tirzepatide activates two gut-hormone pathways (GLP-1 and GIP) instead of one. The advantage holds across the head-to-head trial and matched real-world datasets.

If you're choosing between them with a prescriber, the data points in the same direction across study types: tirzepatide gives more headroom on weight loss, but both drugs require persistence and target-dose escalation to deliver their best results. Cross-study comparisons depend on dose, product, T2D status, and how the study defined persistence.

For help choosing and accessing a provider: Best Zepbound providers that accept insurance · Best Wegovy providers that accept insurance.

How many people actually quit a GLP-1 in the first year?

Why people stop

The Cleveland Clinic and Truveta studies identified four main reasons for stopping, roughly in order of frequency:

1. Cost or loss of insurance coverage. [7][9] Wegovy and Zepbound retail around $1,000–$1,350/month without coverage. Coverage changes can end therapy overnight.
2. Gastrointestinal side effects. Nausea, vomiting, diarrhea, and constipation are common during dose escalation. About 5–10% of trial patients discontinued for GI reasons. [1][2][5][6] Real-world rates are similar when titration is careful, higher when it's rushed.
3. Plateau or perceived non-response. Some patients stop because they think the drug isn't working, even when they're still losing.
4. Drug supply issues. Official shortage status has improved since the 2022–2023 disruption period, but local pharmacy inventory gaps still happen occasionally.

Discontinuation in trials vs real life

So the real-world rate runs roughly 3 to 4 times the trial rate. That's a meaningful difference in who ends up where in the weight-loss distribution.

Restart and switch rates

A piece most coverage misses: discontinuation isn't always permanent. In the Truveta JAMA Network Open analysis, 47.3% of patients with type 2 diabetes and 36.3% without type 2 diabetes reinitiated a GLP-1 within one year of stopping. [9] The strongest predictor of restarting was weight regain after stopping.

In the Cleveland Clinic 2026 follow-up, 27% of patients who stopped switched to a different obesity medication (either between semaglutide and tirzepatide, or to an older anti-obesity drug) within the next 12 months. [8] Real-world treatment is more fluid than trials make it look.

For strategies to maintain coverage continuity, see Does insurance cover Zepbound for weight loss?.

Does real-world GLP-1 results differ if you have type 2 diabetes?

The pattern shows up in every major real-world analysis. JAMA Internal Medicine's matched cohort study found larger weight reductions among adults without type 2 diabetes than among those with diabetes for both semaglutide and tirzepatide. [10] Cleveland Clinic's 2026 follow-up reported that patients with diabetes who stopped GLP-1s actually lost an additional 1.3% in the year after stopping, while patients with obesity but no diabetes regained an average of 0.5%. [8]

• Insurance coverage is different. Type 2 diabetes coverage for GLP-1s is broader and more durable than coverage for obesity alone. This is the biggest driver of the discontinuation gap.
• Some real-world studies use diabetes-labeled products (Ozempic for semaglutide; Mounjaro for tirzepatide) at different dose ranges than the obesity-labeled products (Wegovy; Zepbound). When you compare a result to a study, check the product, the dose, and the population.
• Baseline weight and HbA1c interact with response. Patients with poorly controlled diabetes often see a larger HbA1c improvement and a somewhat smaller percent body-weight reduction than patients with obesity alone.

If you have type 2 diabetes, the trial-vs-real-world question still applies — but the most relevant numbers come from real-world studies that included diabetes patients (Cleveland Clinic, JAMA Internal Medicine, Truveta) rather than from STEP-1 or SURMOUNT-1, which both excluded T2D.

The story the headlines missed: persistence has nearly doubled since 2021

The persistence trend

Why this is happening

• Drug supply has stabilized. The 2022–2023 shortages forced unintended gaps. Those gaps are now less frequent.
• Self-pay pathways have expanded. As of May 2026, NovoCare lists Wegovy pen self-pay at $199/month for the first two monthly fills (through June 30, 2026), then $349/month for 0.25–2.4 mg and $399/month for Wegovy HD 7.2 mg. [23] Eli Lilly lists Zepbound self-pay with $449/month purchase-offer pricing for 7.5–15 mg doses when refill timing requirements are met. [24] The cash-pay floor has dropped meaningfully since 2023.
• Prescribers have learned the titration playbook. Slower escalation, anti-nausea support, and clearer counseling reduce the GI-driven dropouts that defined the early years.
• Coaching programs scaled. Calibrate, Omada, Form Health, Found, and several telehealth services now wrap behavioral support around the prescription. These cohorts persist at 60–70%+. [17][18]

What it means for the narrative

The widely cited “real-world results half of clinical trials” reporting is honest reporting of 2021–2023 EHR data. But that's the early-adopter window — drugs brand new, supply unstable, prescribers still learning titration. The 2024–2026 data is showing what these drugs look like under more settled conditions: persistence rising, weight loss rising with it.

Calibrate's 4-year longitudinal program data (n=37,031 at year 1, n=620 at year 4) is the longest real-world horizon available, and it shows continued average weight loss through year 4. [17] This is company-reported program data — not a randomized post-discontinuation trial — so it shouldn't be read as disproving controlled withdrawal-trial findings. But it does suggest that under sustained support, long-horizon outcomes can stay strong.

Our editorial conclusion: the gap is closing, especially in coverage-stable and coaching-supported cohorts. If you're starting a GLP-1 in 2026, your odds of a trial-like outcome are higher than they were for someone who started in 2021.

The GLP-1 Trial-to-Real-World Translation Score

This framework is original to this page. Use it as a discussion starter with your prescriber, not a personal prediction or treatment recommendation.

How to score yourself

What your score means

Important guardrails: This score is for expectation-setting, not for changing your treatment. It is not medical advice, does not recommend any specific dose, and is not a substitute for guidance from your prescriber. The numbers it points to are anchored to peer-reviewed data, but every person responds differently. Use it to ask better questions, not to make medical decisions.

Example walkthroughs

For a broader medication match tool, try our Find My Path quiz.

Are trial numbers misleading? Efficacy vs effectiveness

The Wegovy and Zepbound trial numbers weren't pharma marketing — they were FDA-required pivotal studies, run to protocols approved before the trials started, with primary endpoints declared in advance. [1][2][5][6] The 14.9% from STEP-1 and the 20.9% from SURMOUNT-1 are real measurements.

But they're measurements taken under a specific set of conditions. Those conditions included:

• Patients who agreed to a year-plus of scheduled study visits
• Free drug supply
• Free dietician counseling at every visit
• A protocol that escalated dose on a fixed schedule
• Trial coordinators following up when patients missed appointments

That's not what most prescribing looks like.

The same efficacy-vs-effectiveness gap shows up across chronic-disease treatment in general — it's a common feature of how medicine actually works once a drug leaves the trial environment.

The mistake isn't trusting the trial numbers. The mistake is using one number to answer the wrong question. If you want to know “what's possible under structured conditions,” look at trial efficacy. If you want to know “what's typical in routine care,” look at real-world effectiveness. If you want a better anchor for your own treatment path, use the Translation Score — as a discussion starter, not as a personal forecast.

Side effects: trial reports vs real-world experience

What the trials reported

From the pooled STEP 1–3 analysis (Wharton et al., DOM 2022) [19]:

For tirzepatide, the Zepbound U.S. label reports GI adverse reactions in 56% of treated patients vs 30% on placebo, with GI-related discontinuation of 1.9%, 3.3%, and 4.3% across 5 mg, 10 mg, and 15 mg vs 0.5% placebo. [6]

What's different in real life

Two things are typically worse in real-world use:

1. Faster titration. Trials follow a slow, careful dose-escalation schedule. Real-world prescribers, especially earlier in the GLP-1 era, sometimes pushed dose increases before patients were ready, which led to more severe GI symptoms.
2. Less anti-nausea support. Trials had research staff available to discuss symptom management. Real-world patients often suffer through it or stop.

In the Cleveland Clinic 2025 study, GI side effects were among the top three reasons for discontinuation. [7] That's why the persistence-doubling we saw from 2021 to 2024 isn't only about coverage — it's also about prescribers learning to titrate more carefully.

Rarer but more serious adverse events

• Gallbladder problems (cholelithiasis): Wegovy trials reported gallbladder-related disorders in 2.6% of patients vs 1.2% on placebo. [5] Rapid weight loss in general increases gallstone risk.
• Pancreatitis: Rare in trials and identified as a postmarketing safety signal. Postmarketing reports can identify safety signals but cannot reliably estimate frequency or prove causality on their own.
• Severe gastroparesis: A small postmarketing signal. Likely related to the same gut-slowing mechanism that gives the drug its appetite-reducing effect.

What happens when people stop GLP-1 medications

What the trials showed

STEP-4 is the headline randomized-withdrawal source. [3] Patients who took semaglutide for 20 weeks and then switched to placebo for 48 weeks regained roughly two-thirds of the weight they'd lost in the first phase. That's the data behind the “you'll have to take it forever” coverage.

The mechanism is straightforward: GLP-1s suppress appetite and slow gastric emptying. Take them away, and the underlying physiology of obesity reasserts itself.

What real-world data shows

Real-world stopping looks different — meaningfully different — because real life isn't a randomized withdrawal.

The Cleveland Clinic 2026 study (Gasoyan et al., Diabetes, Obesity and Metabolism) followed 7,938 adults who started semaglutide or tirzepatide and stopped within 3–12 months. [8] The findings:

• Obesity-treated patients had lost an average of 8.4% before stopping
• One year later, they had regained an average of just 0.5%
• 27% switched to a different obesity medication within 12 months
• 55% had some weight regain; 45% kept losing or stayed the same
• Patients with T2D actually lost an additional 1.3% in the year after stopping

The BMJ/Oxford 2026 meta-analysis

Researchers at Oxford pulled together 37 studies covering 9,341 adults who stopped weight-management medications. [20] Their findings:

• Average regain: about 0.4 kg/month across all weight-loss drugs
• For newer GLP-1s specifically (semaglutide and tirzepatide): about 0.8 kg/month
• Long-term data beyond 12 months for newer GLP-1 drugs remain limited

Why real life diverges from trial withdrawal

1. People don't just stop — they switch. Real-world discontinuation often means moving to another anti-obesity drug, older medications, intermittent dosing, or behavioral programs. [8]
2. People restart. The Truveta JAMA Network Open analysis found 47.3% of T2D patients and 36.3% of non-T2D patients restarted within a year. [9]
3. Real-world stoppers had lost less to begin with. Cleveland Clinic's team noted that real-world stoppers had typically lost about 8% — much less than the 15% trial completers had to lose back. [7][8] Smaller losses are easier to maintain.

What this means for you

• Regain risk is real but the pace is slower in real-world settings than randomized-withdrawal trials suggest
• Maintenance approaches matter: lower-dose continuation, switching, behavioral support
• The biggest predictor of not regaining is having a plan before you stop
• This is exactly the kind of decision that justifies a structured conversation with your prescriber

If a coverage loss is forcing you off a GLP-1, see does insurance cover Zepbound for weight loss and how to appeal a Zepbound denial.

Limitations on both sides

Clinical trial limitations

• Eligibility criteria exclude common real-world patients. STEP-1 excluded patients with diabetes [1]; SURMOUNT-1 had exclusions for several comorbidities and medications. [2]
• Structured follow-up doesn't match routine care. Frequent visits, free drug, dietician counseling, and trial-coordinator outreach are not the norm.
• Dose escalation is protocol-driven. Real-world prescribers escalate based on tolerance, finances, and patient preference.
• Trial timelines may not match patient use. A 72-week trial endpoint is less relevant if a patient stops at 6 months.
• Participants self-select for adherence. Anyone who volunteers for a 68–72 week trial with regular visits is already more committed than average.

Real-world evidence limitations

• Missing or inconsistent weight data. EHR weights aren't measured on a schedule.
• Prescription fills aren't doses taken. Claims data shows what was bought, not what went into the body.
• Care outside the system is invisible. A patient who switches health systems disappears from EHR-based studies.
• Discontinuation is often defined arbitrarily. “60-day gap” or “84-day gap” varies between studies, which makes head-to-head comparison messy.
• Confounding by indication remains. Patients prescribed tirzepatide vs semaglutide may differ in ways the data can't capture.
• Shortages and policy changes distort trends. The 2022–2023 supply problems are baked into the 2021–2023 RWE persistence numbers.

Sponsorship and source-type note

Some of the real-world entries in our comparison table were conducted by health systems (Cleveland Clinic), independent academic groups (Samuels), data companies (Truveta, Optum, Komodo), claims-data firms (Prime Therapeutics), or programs with a commercial interest in GLP-1 patient outcomes (Calibrate, Omada). We've cited the source type for every row so you can weigh it yourself. Studies with industry funding aren't automatically suspect, but they're not equivalent to independent peer-reviewed work either. We've prioritized peer-reviewed, primary-source data wherever it exists.

What we actually verified

Frequently asked questions

Sources

Related guides

• Does insurance cover Zepbound for weight loss?
• Does Medicare cover Zepbound?
• How to appeal a Zepbound prior authorization denial
• Best Zepbound providers that accept insurance (2026)
• Best Wegovy providers that accept insurance (2026)
• Best GLP-1 providers that accept insurance (all medications)
• Best GLP-1 providers that help with prior authorization
• Zepbound Savings Card: how it works, eligibility, and limits
• Cheapest Zepbound without insurance (2026)
• Mounjaro vs Wegovy: head-to-head comparison