Does site rotation actually have evidence behind it, and what kind of reaction warrants stopping?

Two related questions I cannot get straight answers to.

First, site rotation. Everyone says rotate, nobody says on what evidence or at what spacing. I have seen "2 cm", "an inch", "a different quadrant each week" and "a different body region each month" all stated as if they were established facts. For a weekly injection, is any of this actually necessary, or is it advice imported wholesale from insulin practice where people inject four times a day?

Second, thresholds. I now have three firm painless lumps in my abdomen from previous injections, the oldest about four months old and not shrinking noticeably. They do not hurt, they are not red, and they do not bother me except that I am accumulating them. At what point does a local reaction warrant actually stopping the drug rather than adjusting technique? I would like to know the threshold before I am at it rather than after.

On the first question: the direct evidence for site rotation in weekly GLP-1 dosing is essentially absent, and the practice is indeed imported from insulin, where the evidence is good. That import is defensible on mechanism but the recommended spacings are not empirically derived for this use case, and anyone stating "2 cm" as an established figure is quoting insulin guidance. On the second: the threshold for stopping is narrower than people expect, and your nodules are almost certainly not it.

What the insulin evidence actually shows

Lipohypertrophy — firm, rubbery thickening of subcutaneous tissue at repeatedly used sites — is well documented in insulin users, at prevalences commonly reported in the tens of percent among long-term users. The important findings are not that it exists but what it does:

• Absorption from lipohypertrophic tissue is slower and more variable than from normal tissue.
• Injecting into it is associated with worse glycaemic variability and higher insulin requirements.
• Structured rotation and avoiding affected sites reduces its development and improves control.

So the mechanism established in insulin practice is real and it is about pharmacokinetics as much as appearance. That mechanism does not obviously depend on the molecule; it depends on repeated tissue trauma and repeated deposition of fluid in the same place.

Why the transfer to weekly dosing is weak

The dose frequency differs by roughly a factor of 20-30. Someone injecting insulin four times daily delivers about 1,460 injections a year; a weekly agent delivers 52. Lipohypertrophy in the insulin literature is associated with high injection frequency, small site areas and needle reuse, and the incidence scales with all three. At 52 injections a year across even a modest number of sites, the cumulative trauma per unit of tissue is more than an order of magnitude lower.

The trial evidence is consistent with that: injection-site reactions in the pivotal trials of this class run at a few percent, and lipohypertrophy is not a prominent finding. Participants used pens into rotated sites, but nothing in those data suggests a large problem waiting to happen.

So the honest position: rotation is cheap, mechanistically sensible, and consistent with practice in a related context where the evidence is good. It is probably beneficial and it is certainly not harmful. The precise spacing figures are borrowed rather than derived. Sensible practice, without pretending to precision: use more than one region, do not use the same square inch twice in a row, and avoid any site that already has a lump, bruise or reaction in it. That captures the mechanism without inventing numbers.

Your nodules

Firm, painless, non-inflamed subcutaneous nodules persisting for months after injection are a recognised entity, distinct from an inflammatory reaction and distinct from lipohypertrophy:

• Sterile granuloma or fibrotic nodule. A localised foreign-body-type response to injected material, sometimes with a small collection of fibrous tissue. Firm, discrete, painless, and can persist for many months. Not infected. Antibiotics do nothing.
• Lipohypertrophy, which is more diffuse and rubbery rather than discrete and firm, and is associated with frequent injection into the same site.
• Lipoatrophy, the opposite: a depression from localised fat loss, an immunological phenomenon that is much rarer.
• Encapsulated haematoma from an early bleed that organised.

Three discrete firm painless lumps at four months, in the absence of inflammation, most likely fall in the first or fourth category. The relevant practical implications are unglamorous: do not inject into them, because absorption from fibrotic tissue is unpredictable; expect slow resolution over months; and mention them at a routine appointment so they are documented and examined once, because "a lump that is not going away" deserves a look from someone qualified for reasons that have nothing to do with injections.

What warrants stopping

Stopping for a local reaction is rare and the trials bear that out: discontinuation attributable to injection-site reactions is a fraction of a percent in this class. The genuine thresholds:

Stop and seek urgent assessment — these are systemic, not local:

• Any breathing difficulty, wheeze, throat tightness or swelling of lips, tongue or face.
• Widespread urticaria or rash away from the injection site.
• Dizziness, collapse or a sense of impending doom during or after injection.
• Any reaction that is worse and faster on each successive exposure. Escalating reactions are the pattern that precedes anaphylaxis and they are the one thing on this page that should not be watched and waited on.

Seek assessment promptly, and expect a clinician to consider stopping:

• Suspected infection: progressive pain, spreading redness, fever, pus, streaking.
• Severe local reactions at every site that do not respond to any technique or diluent change.
• Ulceration, skin breakdown or necrosis at a site.
• A local reaction accompanied by any systemic symptom at all.

Not a reason to stop, though a reason to change something:

• Mild transient erythema, itch or stinging.
• Bruising.
• Painless nodules, including yours.
• A single vigorous reaction at one site that resolves.

The organising principle: the local-versus-systemic distinction matters more than severity. A dramatic-looking but purely local reaction is usually a technique or formulation problem. A modest local reaction with any systemic accompaniment is a different category. And an escalating pattern across exposures is the one that should not be managed by adjusting your needle length.

edited 7 Aug 2026 by Dr_Otto_Lindqvist — corrected a unit error in the worked example

A table differentiating the reaction types, since the terms get used interchangeably and they have different implications, timescales and responses.

Three practical rules that fall out of the table:

1. Itch means immune or irritant; pain means mechanical or infective. Not infallible, but it is the fastest first cut.
2. Static or improving means benign; progressive means infective. The border-marking test operationalises this.
3. Every site means the material; one site means the injection. This is the discriminator that tells you whether to investigate your vial or your technique, and it is the one people skip.

None of this is a diagnosis and several of the rows are conditions that need examining rather than classifying at home. The purpose of a table like this is to know which questions to answer and which findings are worth reporting, not to replace an examination.

Adding the interaction with anticoagulants and antiplatelet drugs, which explains a large fraction of the bruising reports and is not mentioned in any patient information I have seen for this class.

Subcutaneous injection punctures small vessels. Whether that produces nothing, a pinprick or a 4 cm bruise depends substantially on haemostasis, and a great many people in the population taking these drugs are on something that impairs it:

• Aspirin and other antiplatelet agents, very common in the cardiovascular population, which is exactly the SELECT population.
• Direct oral anticoagulants and warfarin.
• NSAIDs, including intermittent over-the-counter use.
• Selective serotonin reuptake inhibitors, which have a modest antiplatelet effect that is well described and generally forgotten.
• Fish oil at high doses, and some other supplements with a plausible antiplatelet action.
• Alcohol, acutely.

Practical consequences, and the reason to know: bruising in these people is a haemostasis finding rather than a technique failure, so the correct response is not to change needle or angle in search of a fix that does not exist. Direct pressure after withdrawal for thirty seconds or so, without rubbing, does more than any change of equipment. Rubbing the site is the specific thing that turns a pinprick into a bruise and it is what most people do reflexively.

Two other points in the same territory:

• Anticoagulation raises the significance of a haematoma. A large, expanding or painful swelling at a site in someone anticoagulated is worth assessment rather than observation, which is different advice from the same finding in someone not anticoagulated.
• Bruising and infection can look similar at day three, because a resolving haematoma is discoloured and tender. The distinguishing features are the colour evolution, blue-purple then yellow-green, and the absence of heat and of progression.

The reason this belongs in a discussion of injection-site reactions is that the population using these drugs overlaps heavily with the population on antiplatelet and anticoagulant therapy, and a substantial share of what gets reported as an injection-site reaction in that group is ordinary bruising in someone whose blood does not clot as quickly as it used to. Recognising it saves a great deal of pointless technique adjustment.