Skip to content
GLP Loss
← Research
Safety

When the Pharmacy Changes: Why Two Compounded Vials Are Not Interchangeable

Generic substitution works because the FDA makes five findings first. None of them exists for a compounded preparation — and in approved generics, where all five do exist, merely changing a pill's color raised the odds of stopping by 34%.

Hana Brennan9 min read
What makes a pharmacy switch safeEvery guarantee behind generic substitution, and whether it reaches hereGENERICCOMPOUNDEDApproved by the FDAyesnoEquivalence rating existsyesnoneBioequivalence demonstratedyesnoStrength fixed by the approvalyesnoYou are told it changedon the bottleoften notOnly the color changed, and refills stoppedRecoloring a certified-equivalent pill raised nonpersistence 34%A change of pill shape: 66% higher odds of stopping after a heart attack.GLP-1 persistence is 32.3% at one year before any switch happens.

The company that takes the order is generally not the company that makes the drug, and the one that makes it can be replaced between one month and the next. On the approved side of the market that replacement is routine and nearly consequence-free, because a large legal apparatus exists to make it so. None of that apparatus reaches a compounded preparation, and the reason is worth understanding before the second vial arrives.

Substitution is a machine, and it runs on one finding

Virtually every state has laws or regulations encouraging the substitution of drug products, and those laws were written around a federal list. The FDA publishes therapeutic equivalence evaluations for approved multisource prescription products precisely so that fifty states would not each build their own formulary on their own criteria.[1]

The finding behind a swap is specific. Two products are therapeutic equivalents only if they are approved as safe and effective; are pharmaceutical equivalents, containing identical amounts of the identical active ingredient in the identical dosage form and route and meeting applicable standards of strength, quality, purity and identity; are bioequivalent; are adequately labeled; and are manufactured in compliance with current good manufacturing practice regulations.[1] Five findings, all of them made by a regulator, before a pharmacist may hand over a different manufacturer’s bottle.

The FDA states the contrast directly: a generic drug is approved under section 505(j) of the act and must meet all of that section’s requirements, including establishing therapeutic equivalence to a brand name drug, while a compounded drug is not approved by the FDA.[2] A compounded preparation has no Orange Book entry, no equivalence code and no reviewed label. There is no instrument that could declare one pharmacy’s vial interchangeable with another’s, because interchangeability is a finding and nobody is positioned to make it. What the phrase “not FDA-approved” covers in full is set out in the approval article.

What a pharmacy change is permitted to change

Even between approved generics the regulator allows quite a lot of variation. Pharmaceutical equivalents do not necessarily contain the same inactive ingredients, and may differ in shape, scoring, release mechanism, packaging, excipients including colors, flavors and preservatives, expiration date and time, and within limits their labeling.[1] Equivalence survives all of that only because bioequivalence was demonstrated on top of it.

Remove the demonstration and the same list of permitted differences becomes a list of unverified ones. Two compounding pharmacies can differ in the concentration they dispense, in whether the active ingredient is a base or a salt form, in whether a preservative is present at all, in the excipients and any added ingredients, in the beyond-use date assigned to the container, and in the syringe supplied to measure it. Each of those is a documented axis of variation in this market rather than a hypothetical one, and the composition question is the subject of the contents article.

The number on the syringe is not the number in the vial

Concentration is the one that bites hardest, because it is invisible in the instruction a patient is following. An approved vial is dosed by volume against a fixed strength: the Zepbound instructions name a syringe able to measure a specific milliliter volume for a specific dose.[3] That works because the strength behind the volume was set by an approval and cannot move.

A compounded vial’s strength was set by whoever compounded it. If a new pharmacy fills the same prescription at a different milligrams per milliliter, an unchanged habit — the same marks on the same barrel, the same remembered number of units — delivers a different dose. The FDA has documented that dosing errors follow from exactly this class of confusion between milligrams, milliliters and units when patients are handed a multiple-dose vial, and the reported cases are collected in the needle and syringe article. The important point for a switch is structural: nothing about the second vial announces that the first one was different.

The best evidence on what a switch costs came from drugs that were identical

The cleanest measurement of a pharmacy switch was made where the drug did not change at all. A cohort and nested case-control study followed patients discharged after myocardial infarction who started a generic beta-blocker, ACE inhibitor, angiotensin receptor blocker or statin. Across the study, 29% of 11,513 patients saw the color or shape of their pill change. Matching 4,573 episodes of nonpersistence against 19,881 controls, the odds of stopping rose 34% after a change in pill color (adjusted odds ratio 1.34, 95% CI 1.12 to 1.59) and 66% after a change in shape (1.66, 95% CI 1.43 to 1.94).[4]

These are drugs the FDA had certified as therapeutically equivalent. Identical active ingredient, identical strength, identical dosage form, demonstrated bioequivalence, adequate labeling. The only thing that changed was what the tablet looked like, and people stopped taking it.

A companion study in antiepileptic drugs complicates the picture rather than confirming it. Among 11,472 patients who became nonpersistent, matched to 50,050 controls, color discordance preceded nonpersistence with an adjusted odds ratio of 1.27 (95% CI 1.04 to 1.55), and within the seizure-disorder subgroup 1.53 (1.07 to 2.18). But shape discordance came out at 1.47 with a confidence interval of 0.85 to 2.54 — not significant, in the cohort where shape was expected to matter most.[5] Color replicates across both populations; shape does not. Anyone quoting a single number for “what a switch costs” is quoting one cohort.

The baseline it lands on

A disruption matters in proportion to how fragile the pattern already is, and for this drug class the pattern is fragile. Claims data covering 16.5 million commercially insured members identified 4,066 adults with obesity and without diabetes newly starting a GLP-1. Persistence was 46.3% at 180 days and 32.3% at one year. Average proportion of days covered over the year was 51.0%, with 27.2% of members adherent at the conventional 80% threshold, and 11.1% switched GLP-1 drugs.[6]

Two-thirds of that cohort were gone within a year with no pharmacy change involved. An unexplained change in what arrives in the box — a different vial, a different concentration, a different look — is being added on top of that, in a population already leaving at scale.

What a buyer can actually verify, and what they cannot

Some of it is checkable. The FDA publishes the list of facilities registered as outsourcing facilities under section 503B, so a named 503B can be confirmed to be on it.[7] State boards of pharmacy license the 503A pharmacies and hold their disciplinary records. The container that arrives carries its own labeling, including the assigned beyond-use date, and the difference between the two statutory categories is set out in the 503A and 503B article.

Most of it is not. The potency of the particular lot in the particular vial is not published and cannot be inferred from a label. Nor is its sterility testing, its endotoxin result, the source of its active ingredient, or whether a change of supplier happened at all. The FDA puts the structural problem plainly: consumers who purchase compounded drugs online, including through telehealth platforms, may not know the identity of the compounder that produced the drug, including whether it was produced by a compounder whose drugs meet appropriate quality standards and that is appropriately licensed and regulated.[2] The questions that make a pharmacy identifiable at all are in the vetting article.

Why no label answers this

Reading the four approved labels for a substitution instruction returns nothing, and the absence is structural rather than an oversight. Every approved GLP-1 for weight management is a single-source product. No generic semaglutide or tirzepatide has been approved, so no therapeutic equivalence code exists for either molecule, so there is no substitution for a label to govern. The one thing an approved label does supply is the dating and handling regime a container is entitled to, which is a per-product figure rather than a class one — see the storage article.

That silence is the real finding. A buyer looking for the rule that governs a compounder change is looking for a rule that was never written, because the statutory scheme that would have written it applies to approved products and compounded preparations are exempt from the sections that create it.

What this changes about a purchase

It makes the identity of the compounder a term of the transaction rather than a detail. A seller that names its pharmacy has given a buyer something to notice a change against; a seller that does not has removed the possibility of noticing. Compounded drugs are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed, which means the identity of the maker carries more of the weight here than it does anywhere else in a pharmacy.

What a specific vial contains, whether it differs from the last one, and what to do if it does are questions for the prescriber who wrote the prescription and the pharmacist who filled it. Which sellers name a pharmacy at all is recorded in the seller reviews, on criteria set out in the methodology, and the boards begin with compounded semaglutide.

Frequently asked

Is compounded semaglutide the same from every pharmacy?
There is no finding that would make it so. Therapeutic equivalence is a determination the FDA makes about approved products, requiring identical active ingredient and strength, demonstrated bioequivalence, adequate labeling and manufacture under current good manufacturing practice. A compounded preparation is not FDA-approved, has no Orange Book entry and carries no equivalence code, so nothing certifies one pharmacy's vial as interchangeable with another's.
What can actually differ between two compounding pharmacies?
Concentration in milligrams per milliliter, whether the active ingredient is a base or a salt form, whether a preservative is present, the excipients and any added ingredients, the beyond-use date assigned to the container, and the syringe supplied to measure the dose. Even between approved generics the FDA permits differences in excipients, preservatives, packaging and expiration dating — equivalence survives only because bioequivalence was separately demonstrated.
Why does a change in concentration matter so much?
Because the instruction a patient is following is usually a volume or a number of units on a syringe barrel, not a milligram figure. If a new pharmacy fills the prescription at a different strength, the same familiar marks deliver a different dose, and nothing about the new vial signals that the old one was different. The FDA has documented dosing errors arising from confusion between milligrams, milliliters and units with multiple-dose vials.
How much does a pharmacy switch really affect whether people keep taking a drug?
In FDA-approved generics, where nothing about the medicine changes, a change in pill color raised the adjusted odds of nonpersistence by 34% and a change in shape by 66% among patients recovering from a heart attack. In a separate antiepileptic cohort, color replicated at 1.27 but shape did not reach significance at 1.47 (95% CI 0.85 to 2.54). Color is the consistent signal; shape is cohort-dependent.
Can a buyer verify which pharmacy made their vial?
Only partly. The FDA publishes the register of 503B outsourcing facilities and state boards license 503A pharmacies, so a named compounder can be checked. What cannot be checked is the potency of the specific lot, its sterility and endotoxin results, the source of the active ingredient, or whether a supplier changed at all. The FDA notes that consumers buying compounded drugs through telehealth platforms may not know the identity of the compounder.
Do the approved labels say anything about substitution?
No, and the silence is structural. Every approved GLP-1 for weight management is a single-source product; no generic semaglutide or tirzepatide has been approved, so no therapeutic equivalence code exists and there is no substitution for a label to govern. The rule a buyer is looking for was never written, because the statutory scheme that creates it applies to approved products.

Sources

  1. [1] U.S. Food and Drug Administration (2026). Orange Book Preface: Approved Drug Products with Therapeutic Equivalence Evaluations — therapeutic equivalence criteria and state substitution laws (read September 15, 2026) U.S. Food and Drug Administration. Source
  2. [2] U.S. Food and Drug Administration (2025). Compounding and the FDA: Questions and Answers — compounded drugs are not FDA-approved, and consumers may not know the identity of the compounder (read September 15, 2026) U.S. Food and Drug Administration. Source
  3. [3] Eli Lilly and Company (2026). ZEPBOUND (tirzepatide) injection, solution — vial administration and syringe volume instructions (read September 15, 2026) DailyMed, U.S. National Library of Medicine. Source
  4. [4] Kesselheim AS, Bykov K, Avorn J, Tong A, et al. (2014). Burden of changes in pill appearance for patients receiving generic cardiovascular medications after myocardial infarction: cohort and nested case-control studies. Ann Intern Med. PMID 25023248
  5. [5] Kesselheim AS, Misono AS, Shrank WH, Greene JA, et al. (2013). Variations in pill appearance of antiepileptic drugs and the risk of nonadherence. JAMA Intern Med. PMID 23277164
  6. [6] Gleason PP, Urick BY, Marshall LZ, Friedlander N, et al. (2024). Real-world persistence and adherence to glucagon-like peptide-1 receptor agonists among obese commercially insured adults without diabetes. J Manag Care Spec Pharm. PMID 38717042
  7. [7] U.S. Food and Drug Administration (2026). Registered Outsourcing Facilities under section 503B of the Federal Food, Drug, and Cosmetic Act (read September 15, 2026) U.S. Food and Drug Administration. Source

More in Safety