The CVS Caremark Forced Switch: What Happened to 72,245 Zepbound Patients
Only 43% did what the formulary intended. The rest retained, worked around, or stopped treating entirely.
The Conventional View
The mechanics of formulary exclusion are well understood. The downstream patient-level outcomes at scale are not. CVS Caremark tested this at scale on July 1, 2025 — removing Zepbound from its standard, Advanced Control, and Value formularies covering an estimated 25-30 million lives. Wegovy became the preferred next-generation GLP-1 for obesity.
We tracked 72,245 stable Zepbound patients (2+ fills in H1 2025 at CVS-adjudicated plans) through eight months of post-switch claims data. The question: did the patients comply, and did the substitute therapy hold them?
Where 72,245 Zepbound Patients Actually Went
The finding: less than half of stable patients followed the intended path. A third were never truly disrupted — consistent with employer-specific custom formularies or successful medical exceptions — reframing the actually disrupted population as approximately 48,853 patients. Among those who were disrupted, more found workarounds or exited therapy than complied with the switch.
The Persistence Deficit: Forced Substitution Is Not Outcome-Neutral
The 14.1 percentage point deficit between forced switchers and undisrupted controls decomposes into two additive effects: approximately 7.8pp attributable to the drug switch itself (Zepbound inherently outpersists Wegovy even among voluntary starters — a product-level finding) and approximately 6.3pp attributable to the act of coerced substitution (measured on the same drug, same payer, same time period — a behavioral finding). The gap widened from 90 to 180 days, consistent with progressive attrition among patients never fully engaged with the substitute therapy.
The Clinical Mechanism: Subtherapeutic Restart Dosing
A partial explanation for the persistence deficit: 53.6% of all forced switchers were started at Wegovy 0.5mg or below — the earliest titration steps, requiring months to reach maintenance dosing. Among patients on the highest Zepbound doses (10-15mg, includes 12.5mg titration step), a third were restarted at 0.5mg, requiring approximately 16 weeks of titration to reach maintenance dosing. There is no FDA-established dose equivalence between tirzepatide and semaglutide. Prescribers were left to improvise.
The clinical experience this creates is the inverse of a normal Wegovy start. An organic starter experiences progressive benefit from baseline — each dose increase brings more effect. A forced switcher experiences regressive benefit from a therapeutic tirzepatide dose — weeks of subtherapeutic semaglutide before reaching maintenance. The same drug. A fundamentally different patient experience.
They're Still Diagnosed. They've Stopped Treating.
Among the approximately 10,200 patients who exited all GLP-1 therapy after the formulary change, 76% remain medically active — filling other prescriptions, seeing physicians. Among those still active:
| Metric | Value |
|---|---|
| Still carry E66.x obesity diagnosis | 43.3% |
| Still seeing original Zepbound prescriber | 29.9% |
| Filled any other prescription (months 3-6) | 76.2% |
The clinical indication persists. The treatment has stopped. This is consistent with formulary-driven treatment abandonment rather than clinical resolution — though some share of exiters may have transitioned to cash-pay channels (LillyDirect at $349-499/month) not visible in adjudicated claims. Cash-pay options were less widely adopted in mid-2025 than they are today, but this limitation cannot be fully quantified.
CVS Is Still Paying for Tirzepatide
Approximately 10,800 former Zepbound patients filled Mounjaro — the diabetes-branded form of the same tirzepatide molecule — after the exclusion. This appears to be a prescriber-led indication workaround to maintain tirzepatide access through the diabetes formulary pathway.
| Metric | Value |
|---|---|
| Former Zepbound patients filling Mounjaro post-switch | ~10,800 |
| Without any T2D diagnosis in claims history (since Jan 2024) | 87.8% |
| With documented obesity diagnosis within 120 days | 52.8% |
| Mounjaro fills adjudicated through CVS | 94.6% |
At the transaction level, CVS replaced a $1,032 tirzepatide fill with a $1,287 semaglutide fill — and is simultaneously adjudicating ~9,500 tirzepatide fills under a different brand name. Combined obesity GLP-1 paid spend at CVS increased approximately 15% from Q2 to Q4 2025 (at transaction level; net-of-rebate economics are not observable in claims). The Mounjaro workaround compounds this: CVS excluded tirzepatide for obesity and is now adjudicating tirzepatide for approximately 9,500 patients without documented diabetes. The cost savings rationale for the formulary change depends entirely on confidential manufacturer rebates not observable in claims data.
The data does not support the assumption that forced therapeutic substitution at PBM scale is outcome-neutral. Among the 48,853 patients who were truly disrupted:
- 64% switched to Wegovy at CVS — but persisted at rates 6.3pp below organic starters on the same drug
- 14% exited GLP-1 therapy entirely while remaining medically active with documented obesity
- 10% maintained tirzepatide through an indication workaround that creates audit exposure for prescribers and cost leakage for the payer
- Transaction-level pharmacy costs increased, not decreased
These findings are relevant to any manufacturer, payer, or PBM evaluating formulary exclusion as a cost lever for high-utilization therapeutic classes. With orforglipron (Eli Lilly's oral GLP-1) facing an FDA decision on April 10, 2026, the next round of formulary positioning decisions is imminent. The downstream cost of disruption — in patient outcomes, prescriber behavior, and unintended workarounds — may exceed the rebate economics that motivate it.
Methodology
- Open medical + pharmacy claims on 330M+ lives (PurpleLab)
- Formulary data provided by MMIT
- Cohort: GLP-1 pharmacy claim + E66.x/E11.x/Z79.85 diagnosis, March 2023-March 2026
- Stable Zepbound patients defined as 2+ fills at CVS-adjudicated plans in H1 2025
- CVS identification via payer name column (lower bound — true disrupted population likely larger)
- Persistence: refill activity at 90 and 180 days post-index
- Three-arm comparison: forced switchers at CVS, organic Wegovy starters at CVS (same drug, same payer, same period), undisrupted Zepbound at non-CVS payers (same drug, different payer, same period)
Limitations
- Cash-pay channels (LillyDirect, oral Wegovy self-pay) are invisible in adjudicated claims and likely overstate the treatment exit count
- Confidential manufacturer rebates cannot be observed — the net-of-rebate cost picture will differ from the transaction-level analysis presented here
- January-February 2026 data shows confirmed claims processing lag; December 2025 is treated as the last fully reliable month
- Organic Wegovy starter prior GLP-1 history was not verified — the 6.3pp forced-switch effect should be interpreted as consistent with, rather than conclusive evidence of, a behavioral penalty
- CVS payer identification is a lower bound
- Observational design — patients may differ across arms in ways not captured by claims
This is independent research. No manufacturer funded this analysis.
We can run this analysis for your specific brand, payer set, and competitive landscape.
Book a Consultation →