Red, raised and itchy at the injection site for four days. How do I tell a local reaction from an infection?

Four days ago I injected into the left side of my abdomen. Within an hour there was a coin-sized red area, slightly raised, and it itched. It is now day four: the redness is roughly the same size, maybe slightly larger, still itchy rather than painful, warm to the touch but not hot, and there is a firm lump under it that I can feel but not see.

My previous sites have all done something mild for a day or so and then resolved. This one has not. I have no fever and feel completely well.

What I cannot work out is whether this is a more vigorous version of an ordinary local reaction or the beginning of an infection, and I do not want to be the person who goes to a doctor about a bruise or the person who ignores cellulitis for a week. What are the features that actually distinguish them, and is there a timing pattern that helps?

Details in case they matter: 29-gauge insulin syringe, reconstituted with bacteriostatic water, vial about two weeks old and kept refrigerated, swabbed the stopper and the skin with alcohol and let both dry.

On injection depth, since the accepted answer flags it as the likely variable and it is the most common technical cause of avoidable local reactions.

Why depth matters

Three tissue layers, three quite different responses to a bolus of fluid:

• Intradermal (within the dermis, roughly 1-2 mm deep). Dense in mast cells, dendritic cells and nerve endings, and mechanically tight. A bolus here produces a visible raised bleb, immediate itch and stinging, and a wheal. It also alters absorption. This is the layer you produce by injecting at too shallow an angle, or by using a very short needle into a pinched fold and going through the far side.
• Subcutaneous (adipose, from roughly 5 mm down depending on site and body composition). Loose, well perfused, few mast cells relative to dermis, and this is the intended target. Reactions here are usually mild.
• Intramuscular (deeper still). Faster absorption, more pain, more bleeding, and for a formulation designed for subcutaneous depot absorption it changes the pharmacokinetics. Not dangerous as a one-off but not the intent.

Reported technique variables that produce shallow deposition:

• Angle. A 4-6 mm needle is designed to go in perpendicular. Angling it to 45 degrees, which people do out of caution, reduces effective depth by roughly 30% (a 6 mm needle at 45 degrees reaches about 6 × sin 45° = 4.2 mm). With a longer needle a shallow angle is the standard technique; with a short one it puts you in the dermis.
• Pinching a fold with a short needle. Pinching is intended to lift subcutaneous tissue away from muscle, which is useful with a longer needle in a lean person. With a 4-6 mm needle it can thin the fold enough that you traverse it.
• Not inserting fully. Hesitating and leaving the hub 2 mm proud of the skin converts a 6 mm needle into a 4 mm one.
• Site. Subcutaneous thickness varies enormously between abdomen, thigh, upper arm and flank, and between people. A needle length that is comfortably subcutaneous on the abdomen may not be on a lean thigh.
• Withdrawing too fast. Leakage back along the needle track deposits fluid in the dermis and on the skin. A few seconds' pause before withdrawal reduces both leakage and the loss of dose.

Other technique variables with a plausible link to local reactions

• Injecting cold. Cold fluid stings and produces more local vasoreaction. Allowing the syringe to reach room temperature for a few minutes is widely reported to reduce it, and it is one of the few mitigations with a straightforward physical rationale.
• Injection speed. A rapid bolus distends tissue mechanically. Slower delivery over five to ten seconds is consistently reported as more comfortable.
• Volume. Larger volumes into one site produce more distension and more reaction. This is a real argument against very dilute reconstitution when it results in large injection volumes.
• Alcohol not dry. Injecting through wet alcohol drags it into the puncture and stings substantially. Let it evaporate.
• Needle reuse. Blunts the tip, increases tissue trauma, increases bruising and lump formation, and is a sterility problem. There is no upside.
• Injecting into an existing lump, bruise or nodule. Worse reaction, unpredictable absorption. Sites that already have something in them should be left alone.

None of the above is an instruction to inject anything. It is a description of the technique variables that determine which tissue layer receives a subcutaneous injection, which is the mechanism behind most of the local reactions people report.

Adding the incidence context, because the question implicitly asks how unusual this is.

Injection-site reactions in the obesity and diabetes trials of this class run at low rates, generally a few percent, and are consistently higher than placebo — which is notable given that the placebo arms were also injecting weekly. In SURMOUNT-1 injection-site reactions were reported in the region of 5-8% across the tirzepatide arms against roughly 2-3% on placebo [1]. Semaglutide trials report them at similarly low single-digit rates [2]. Discontinuation attributable to injection-site reactions is rare in all of them.

Two comparisons that put those figures in perspective:

• Exenatide extended-release produced injection-site nodules at rates an order of magnitude higher, commonly reported in the tens of percent, because it was a microsphere depot formulation designed to sit in the tissue and release slowly. That is a formulation effect, and it demonstrates that the injectate matters far more than the act of injecting.
• Insulin, injected many times more often, produces lipohypertrophy — firm fatty thickening at repeatedly used sites — in a substantial fraction of long-term users, and it measurably alters absorption. That is the best evidence that site rotation matters, and it comes from a population injecting several times a day for decades rather than once a week.

Why the trial figures probably understate what people encounter outside trials:

• Presentation. Trial participants used prefilled pens with fixed short needles designed for the purpose. Reconstituting a lyophilised vial and drawing with an insulin syringe introduces variables that do not exist with a pen: reconstitution technique, diluent choice, concentration, injection volume, needle length and depth control.
• Diluent. The commercial formulations have defined excipients and a defined pH. Reconstituted material has whatever diluent was used, which is a variable the trials did not have.
• Ascertainment, the same limitation as everywhere else in these safety tables. Nobody was asked weekly about their injection sites.

The practical upshot: an isolated mild local reaction is unremarkable and common. Reactions at every site, worsening over time, or reactions that are severe are worth investigating, and the highest-yield thing to investigate is the injectate and the technique rather than the molecule, because the molecule is the part that has been through trials in tens of thousands of people with a low reported rate.