Patients bring me lab reports more often than they bring me food logs. Usually the report came from a panel they bought online, sometimes from their prescriber, and almost always with a column of arrows down the right margin that has been worrying them for a week. The question is nearly the same every time. Is this the medication, is it the weight loss, or is it something I should be frightened of.
Most of the time it is the second one, and that is worth understanding rather than simply being reassured about. A GLP-1 changes two things at once. It changes body composition, and it changes how much food goes into the body every single day. Blood work sees both of those at the same time and cannot tell you which one it is looking at. That is the actual skill in reading these panels. Not knowing the reference range, which is printed right there on the page, but knowing what else could produce that number in a person who has been eating considerably less for four months.
Glucose, and the trap inside A1c
Glycemic markers are the ones patients watch most closely, and they are usually the ones that bring the best news. Hemoglobin A1c reflects roughly the last two to three months of blood sugar exposure, and the American Diabetes Association's 2026 Standards of Care places the prediabetes range at 5.7 to 6.4 percent, with impaired fasting glucose at 100 to 125 mg/dL. Watching a number move down through that band over two draws is one of the more satisfying things in this work.
Here is the part that is less often explained. A1c is not a glucose measurement. It is a measurement of how much glucose has stuck to hemoglobin, which means it depends on how long your red blood cells have been alive to collect it. Anything that changes red cell turnover moves the result without any change in blood sugar at all. Iron deficiency is the version of this most relevant to people eating much less than they used to, and the literature there is genuinely mixed. Several studies report that iron deficiency pushes A1c upward and that correcting it brings the number back down, others find the opposite or no difference, and the mechanism is still argued about. What everyone agrees on is the practical instruction: near a diagnostic threshold, in someone whose red cell indices are not normal, A1c alone should not settle the question. Fasting glucose, and where the picture is unclear a tolerance test, are what settle it.
I also watch fasting insulin alongside glucose in the people for whom it is appropriate, because glucose can hold steady for years while the amount of insulin required to hold it there climbs. That is the change that tends to arrive first, and it is invisible on a panel that only measures the sugar.
The markers that fall because you are simply eating less
This is the category that gets missed, and it is the category I care about most, because it is the one where nutrition actually changes the outcome rather than just explaining it.
When appetite drops sharply, total intake drops with it, and micronutrients fall in rough proportion to the food that is no longer being eaten. Iron, vitamin B12, folate and vitamin D are the ones I follow most often, partly because deficiency in any of them produces fatigue that patients will otherwise attribute to the medication, or to the weight loss, or to themselves. Fatigue at month five is a symptom with a long list of causes, and several of the items on that list are correctable in a few weeks once you know which one you are dealing with.
Ferritin deserves a specific warning because it is the marker most likely to be read wrong. Ferritin is a reasonable proxy for iron stores in a healthy person, but it is also an acute phase reactant, which means it rises with inflammation regardless of how much iron is actually in the body. In the presence of inflammation, serum ferritin stops tracking iron availability, and a normal-looking ferritin can sit on top of a real deficiency. This is why a ferritin drawn on its own is a weak test, and why it should be read next to an inflammatory marker and the rest of the iron studies rather than alone.
Protein intake is the other thing this category is quietly measuring. The standard protein intake figure most people have heard, 0.8 g per kilogram of body weight, was set for maintenance in generally healthy adults, not for someone losing weight rapidly. In practice I work toward roughly 1.6 g per kilogram during active loss, delivered as 25 to 40 g at each main meal rather than concentrated at dinner. The reason is the body composition data. In the 2025 tirzepatide body composition substudy, about 26 percent of the weight lost was lean mass, and it is worth saying in the same breath that the placebo group in that trial was at 25 percent, because the honest version of this concern is that it is largely what weight loss does rather than something the drug does on its own. A 2026 systematic review across thirty-five studies put the median at 28.3 percent. No blood test measures muscle. What the panel can tell you is whether the eating pattern underneath is thin enough that other things are falling too.
Lipids, liver, and things that move for more than one reason
Lipids usually improve with meaningful weight loss, and I would rather follow apolipoprotein B than a standard cholesterol panel where it is available, because it counts the actual number of atherogenic particles instead of estimating the cholesterol carried inside them. Liver enzymes commonly improve as well, particularly in people who arrived with fatty liver. When they move the wrong way during rapid loss, that is a finding to bring to the prescribing clinician promptly rather than to a nutrition appointment three weeks out, because gallbladder disease is more common during fast weight loss and it presents through this part of the panel.
None of these numbers should be read as a single snapshot. A first draw is a starting position. It only becomes information on the second one.
Why the panel you bought online often misleads
Direct-to-consumer testing has done something genuinely useful, which is to make a blood draw easy to obtain for people who could not otherwise get one. The failure is not in the assay. Commercial laboratories run good assays.
The failure is in the three things around the assay. Which markers were selected, given your medications and your symptoms and what you were actually trying to learn. When the sample was drawn, which matters enormously for hormones and meaningfully for glucose and iron. And what the result means in a person with your particular history, as opposed to in the reference population the range was built from. A panel that arrives as a list of arrows with no one attached to interpret it tends to produce two outcomes, both bad: worry about a value that is normal for you, or reassurance about a value that is not. The blood draw is the cheapest part of this.
What to ask your clinician to follow
If you are on a GLP-1 and want a short, reasonable list to raise at your next visit, I would ask about glycemic markers including fasting insulin where appropriate, a complete blood count with iron studies read together rather than a lone ferritin, vitamin B12 and vitamin D, a metabolic panel including liver enzymes, and a lipid assessment. Ask when the draw should happen relative to your dose and your last meal, because that answer changes some of these numbers more than anything you ate that week. And ask what the plan is for the repeat, because the second draw is where the information lives.
This is general education rather than individual medical advice. Which markers belong on your panel, how often, and what your results mean are decisions for you and your own clinical team.
New essays go out here on Tuesdays and Fridays. The full archive is at https://vitaearete.substack.com/archive. And when you want a plan built around your own labs, medications, and life: https://vitaearete.com
This article is general nutrition education, not individualized medical or nutrition advice, and it does not create a dietitian–client relationship. Medications and their side effects should be managed with your prescribing clinician. See the full disclaimer.