Companion data & citations

The Tirzepatide Persistence Gap — Sources & Data

Every figure in the scrollytelling (tirzepatide-persistence.html) is reproduced here with its source and tagged OBSERVED (verifiable from primary sources) or MODELED (constructed from stated assumptions, anchored to observed priors).

The honest core. The five-year retention curves, lifetime values, the $69B exposure and the $24–35B addressable wedge are modeled — built by exponential decay from two published endpoints and multiplied across an inferred patient count. The direction is robust; the precision is not. This is a strategy estimate, not a forecast.

By Matt East. For entertainment purposes only — analysis reflects the author's personal views and does not represent any employer or affiliated organization. Not investment advice.

0. Scene map

Movement	Scenes	Data §
The Exposure	A $69B leak	§7
The Gap	Trial vs. real world · It decays · Obesity breaks faster	§1, §2
The Cause	Why they quit · Friction not biology · Two readings	§3
The Unit	One refill · $5,980 a year	§4
The Model	Two anchors one curve · Area is the money	§5
The Gap in $	Deflate the gap	§6
The Scale	×5.5M · Most of it isn't yours	§7
The Range / Inversion / Answer	Sensitivity · Bear case · Operating system	§8, §9, §10

1. The persistence gap OBSERVED

Discontinuation within ~1 year. Trial endpoint is 72 weeks; real-world windows are 6–12 months, so a matched-window gap would be larger.

Cohort	Window	Discontinuation
SURMOUNT-1 (trial)	72 wk	15.1%
Optum obesity	6 mo	44.6%
Truveta T2D	12 mo	46.5%
Academic clinic	12 mo	50.0%
Truveta obesity	12 mo	64.8%
Real-world median	—	~50%

2. Retention over time & indication split OBSERVED / MODELED

12-month retention is observed; the 12–24 month segments are extrapolated (only 12-month indication-specific points are directly published).

Series	12-mo retention	vs. trial
Trial (SURMOUNT-1)	86%	baseline
Real-world T2D	54%	−32 pp
Real-world obesity	35%	−51 pp

Truveta data span 2018–2023 and include liraglutide-era users; tirzepatide-specific retention is likely somewhat higher.

3. Why patients discontinue OBSERVED

Cleveland Clinic decomposition, non-T2D obesity (n=288), stated reason at point of discontinuation.

Reason	Share	Bucket
Cost / insurance	47.6%	commercial
Side-effect intolerance	14.6%	drug
Unable to fill (shortage)	11.8%	supply artifact
Unspecified	11.1%	—
Other	10.8%	—
Switched to compounded	2.4%	supply artifact
Unsatisfactory weight loss	1.7%	drug

~16% is drug-related (AE + poor response); ~14.2% (shortage + compounded) are 2022–2024 supply artifacts. For contrast, Truveta extracted cost mentions from clinical notes at only ~13% — stated reason vs. note mention diverge by method.

4. Unit economics MODELED (illustrative)

Per 28-day fill (patient-month)	Value
Net revenue	$460
Gross profit (~82%)	$377
Operating profit (~41%)	$189
Net revenue / fully persistent year (×13)	$5,980

$460 is a 2026 base-case estimate; net realized US price disclosed −13% YoY in Q1 2026 (prior $520 anchor was 2024-era). 41% operating margin is company-level; tirzepatide-specific is likely higher.

5. Five-year retention construction MODELED

Two observed anchors + a constant-per-patient-month hazard (exponential decay), S(t)=e^−λt.

Anchor	Observed point	λ / PM	Cumulative PM (5-yr)
Trial (SURMOUNT-1)	84.9% @ 18 PM	0.0091	~49
Real-world obesity (Truveta)	35% @ 13 PM	0.0808	~12
+ Reinitiation (35% restart)	—	—	+2.75

6. LTV & the closeable gap MODELED

Per obesity patient (5-yr)	Value
Trial-like LTV (49 PM × $460)	~$22.5K
Real-world LTV (12 PM × $460)	~$5.5K
Naive gap	$17.0K
− Reinitiation recapture	−$1.3K
− Competitive switching (placeholder)	−$1.5K
Closeable gap	~$14K

T2D LTV gap is smaller (~$10K) given higher retention (~54% at 13 PM) and higher reinitiation (~51% within 1 yr).

7. Enterprise exposure & addressable wedge OBSERVED revenue / MODELED exposure

Step	Value
2025 US tirzepatide revenue ($13.5B Zepbound + $13.6B Mounjaro) OBS	$27.1B
Est. US unique patients (back-solved) MOD	~5.5M
Indication-weighted closeable gap (65% obesity × $14K + 35% T2D × $10K) MOD	~$12.5K
Lifetime revenue exposure (5.5M × $12.5K) MOD	~$69B
Operating-profit exposure (× 41%) MOD	~$28B
Lilly-addressable (35–50% of leak) MOD	$24–35B

Lilly owns: affordability, copay, LillyDirect, titration support. Not Lilly's: prior authorization, formulary placement, Medicare obesity coverage (PBMs, payers, CMS).

8. Sensitivity MODELED (illustrative, not a forecast)

Lifetime revenue captured ($B) from one 2025 US cohort, by closure rate × net revenue per patient-month (100% row spans the $24–35B addressable wedge).

Closure	$380 (stress)	$460 (base)	$550 (bull)
25%	$5.8B	$7.1B	$8.5B
50%	$11.7B	$14.1B	$16.9B
75%	$17.5B	$21.2B	$25.4B
100%	$23.4B	$28.3B	$33.8B

Base-case operating profit recaptured @ 41% margin: 25% → ~$2.9B, 50% → ~$5.8B, 75% → ~$8.7B, 100% → ~$11.6B.

9. Bear case

Net price decline swamps gains. A further 15–20% net-price cut over 2026–2028 could make closing 50% of the gap net to flat revenue.
Reinitiation already recaptures the leak. 35–51% of patients restart within 12 months; modeling discontinuation as permanent overstates the gap.
Pre-2026 supply was the binding constraint. 2023–mid-2025 lost persistence ≠ lost revenue — Lilly was capacity-rationed. The thesis is forward-looking only.
Orforglipron cannibalizes injectable tirzepatide. Oral GLP-1 preserves franchise persistence but at lower per-fill economics — margin compression.
Cost-effectiveness threshold sits below current net price. ~30% additional price cut needed to hit $100K/QALY (JAMA Health Forum, Hwang 2025).
Acquisition is still the dominant growth lever — today. Q1 2026 US Zepbound grew 79% YoY. Persistence is the quality-of-revenue question, not yet the rate-of-growth question.

10. The adherence operating system

Lever	~% of leak	Lilly ownership
PA friction / coverage churn	25–30%	Influences
Patient out-of-pocket affordability	15–20%	Partially owns
AE management / titration	~15%	Influences
Patient engagement / habits	~10%	Influences
Fill failures / shortages	12% → ~2%	Owns
Portfolio recapture	—	Owns

11. The OBSERVED / MODELED ledger

OBSERVED — verifiable from primary sources

SURMOUNT-1 trial completion at 72 weeks: 84.9% — NEJM 2022 (Jastreboff et al.).
Real-world obesity 12-month retention: 35% — Truveta; JAMA Network Open 2025 (Rodriguez/Emanuel).
US tirzepatide FY2025: $27.1B ($13.5B Zepbound + $13.6B Mounjaro); global $36.5B — Lilly Q4 2025 8-K (Feb 4, 2026).
Cleveland Clinic discontinuation drivers (n=288): 47.6% cite cost/insurance — Gasoyan et al., Obesity 2025.
Real-world median 12-month discontinuation: ~50% — Optum Market Clarity, Truveta, academic clinics.
Lilly US manufacturing commitment: >$50B since 2020 — Lilly press release, Feb 26, 2025.

MODELED — constructed from simplifying assumptions

~5.5M US patients touched in 2025 — back-solved from $27.1B ÷ ~$460/fill × ~10.7 fills/yr. Plausible range 5.5–7M against IQVIA Q2 2025 cross-indication GLP-1 pool of ~11M.
5-year retention curves — exponential decay anchored to observed endpoints (λ_trial=0.0091/PM, λ_RW=0.0808/PM); cumulative trial ~49 PM, RW obesity ~12 PM, +2.75 PM reinitiation.
~$14K closeable obesity LTV gap (~$10K T2D) — naive 5-yr gap × $460 net/fill, less reinitiation (~$1.3K) and competitive switching ($1.5K placeholder).
$69B / $28B / $24–35B — exposure, operating-profit exposure, and Lilly-addressable wedge from the per-patient gap × patient count × margin × addressable share.

12. Primary sources

Jastreboff AM, et al. Tirzepatide Once Weekly for the Treatment of Obesity (SURMOUNT-1). NEJM 2022;387:205–216.
Rodriguez P, Emanuel E, et al. Real-world GLP-1 persistence/retention (Truveta). JAMA Network Open 2025.
Gasoyan H, et al. Discontinuation of anti-obesity medications (Cleveland Clinic, n=288). Obesity 2025.
Samuels S, et al. Real-world persistence. Diabetes, Obesity & Metabolism (DOM) 2025.
Truveta Research, ISPOR 2025 (clinical-note cost mentions ~13%); Optum Market Clarity.
Hwang T, et al. Cost-effectiveness of tirzepatide. JAMA Health Forum 2025.
Eli Lilly Q4 2025 8-K (Feb 4, 2026) and Q1 2026 disclosures; manufacturing press release (Feb 26, 2025).
IQVIA, cross-indication GLP-1 patient pool, Q2 2025.

13. Methods & caveats

Cross-window comparison. Trial discontinuation is at 72 weeks; real-world bars are 6–12 months. At matched windows the gap is larger, not smaller.
Decay assumption. Constant per-patient-month hazard is a simplification; real hazards are front-loaded, which would lower early-month retention further.
Patient count. ~5.5M is back-solved and sits at the low end of a 5.5–7M plausible range.
Liraglutide-era contamination. Truveta 2018–2023 data include earlier GLP-1s; tirzepatide-specific retention is likely higher.
Dollar figures are illustrative. Exposure and capture ranges are strategy estimates to size the opportunity, not formal forecasts.

14. Disclaimers

For entertainment purposes only. This is independent editorial analysis of public data, reflecting the author's personal views. It does not represent any employer or affiliated organization.

Not investment advice. Nothing here is a recommendation regarding any security or company.

Not medical advice. Persistence and discontinuation are clinical decisions for patients and clinicians.