The nutritional argument about these drugs is almost always made backwards. It starts from a mechanism — people eat much less, therefore they must take in less of everything — and arrives at a list of vitamins and minerals, as though the list had been measured. It has not been. What follows separates the arithmetic that is solid from the deficiency figures that do not exist, and names the literature the missing figures are being borrowed from.
The one part nobody disputes
Intake falls, and it falls a long way. A 2026 meta-analysis of standardized test meals pooled four arms from three placebo-controlled trials covering 209 participants and found a treatment difference of −1,132 kJ (95% CI, −1,449 to −815), roughly 271 kcal at a single free-access lunch, with no heterogeneity between studies.[1] A separate synthesis of 19 randomized trials across four drug development programs put the decline in daily energy intake at 24.0% to 39.2% across drug classes, with model-estimated deficits reaching 1,200 kcal a day.[2]
That is the premise and it is not in dispute. It is also where the eating article takes over: what the trial protocols prescribed, where the protein numbers come from, and why a gram target sold as validated for this population is borrowed from a different literature. This page is about what happens to everything on the plate that is not protein or calories.
Two instruments, two answers, one set of trials
The same 19-trial synthesis did something unusual: it compared what the trials reported against what the trials had measured. Malnutrition reported by investigators as an adverse event occurred in 0.12% of participants. Objective laboratory screening in the same programs found a total lymphocyte count below 910 per microliter in 2.90% of participants on active therapy, against 1.77% on placebo.[2]
Read those two numbers next to each other. A blood test found the marker roughly twenty-four times as often as a clinician wrote the word down, and the authors drew the obvious conclusion: standard adverse-event reporting understates nutritional risk in this setting. The same analysis put fat-free mass loss on tirzepatide 15 mg at 1.60 kg, about 2.80% of body weight — a figure that belongs to the lean-mass article and is repeated here only because it came out of the same pooling.
One caution belongs with the lymphocyte figure. A low total lymphocyte count is a screening marker used in nutritional assessment, not a diagnosis of malnutrition and not a measurement of any vitamin or mineral. It is the strongest objective signal these trial programs produced, and it is still a proxy.
What has actually been measured in people taking the drug
Very little, and the best available measurement is small. A prospective observational study followed adults with obesity starting semaglutide or tirzepatide for 24 weeks in a real-world clinic with no structured nutrition intervention, and analyzed 28 of them. Total energy intake fell significantly with parallel reductions across macronutrients, and there were significant declines in intake across multiple micronutrients. Diet quality did not improve. Higher absolute protein intake tracked greater preservation of estimated skeletal muscle mass (r = 0.41, P = 0.0498), and roughly a quarter of the weight lost was estimated skeletal muscle.[3]
The authors call the results exploratory and hypothesis-generating, cite the modest sample and the attrition, and say larger adequately powered studies are needed.[3] Those caveats are load-bearing. Twenty- eight people with food records is the state of the direct evidence on micronutrient intake in this population, and food records measure what goes in rather than what ends up in the blood.
The gap may start before the prescription does
The population being prescribed these drugs is not nutritionally average to begin with, and one analysis measured that directly. A cross-sectional study of National Health and Nutrition Examination Survey data from 2005 to 2020 covering 16,143 adults, representing 97.7 million, found 43.5% (95% CI, 42.6% to 44.5%) met the prescribing criteria for these medicines.[4]
Those eligible adults scored worse on diet quality — a Healthy Eating Index of 49.2 against 51.7 (P < 0.001) — with poorer component scores for fruits, vegetables, whole grains and plant proteins, and modestly higher inadequacy for vitamin C, vitamin A, magnesium, vitamin E, vitamin D, fiber, potassium and vitamin K.[4] The difference largely disappeared after adjustment for waist circumference, which the authors read as central adiposity accounting for most of the association.[4]
This is the correction most coverage skips. A person who starts one of these drugs with a thin micronutrient margin and then eats a third less is starting from a worse place, and the drug is not the whole explanation for where they end up. It also means that a before-and-after measurement is the only design that could settle the question, and almost nobody has run one.
Which nutrients, and on what basis
Reviews of this question converge on a similar list and are careful about its status. A 2026 clinical review synthesizing dietary studies, observational cohorts, mechanistic work and pharmacovigilance reports named iron, vitamin B12, vitamin D, calcium, magnesium and zinc as the most relevant signals, with thiamine, folate, vitamin A, other fat-soluble vitamins and potassium as context-dependent concerns.[5]
Its own summary of the evidence is the sentence worth carrying: most abnormalities reported to date are subclinical or indirect, with clinically meaningful consequences possible in susceptible people — those with prior bariatric surgery, gastrointestinal disorders, poor baseline diet quality, older age, or prolonged nausea and vomiting — and prospective studies still needed to define incidence and clinical relevance.[5] A narrative review published the same year reported that observational data describe deficiencies developing within twelve months, most commonly vitamin D followed by thiamine and other B vitamins, with iron, calcium, magnesium and potassium also observed.[6]
Notice what kind of evidence that is. Observational data, mechanistic plausibility and case reports. None of the 19 randomized trials pooled across the four development programs reported serum micronutrient status as an outcome, and the review above closes by saying prospective studies are needed to define incidence.[5] Anyone quoting a percentage of users who become deficient in something is quoting a figure that has not been generated.
Why the framework is borrowed from bariatric surgery
The monitoring advice attached to these drugs comes almost entirely from a different intervention. Metabolic and bariatric surgery has mandatory, structured nutritional monitoring built into its guidelines, because it combines restriction with measurable malabsorption. A 2025 review set the two side by side and found that no equivalent framework exists for this drug class, calling that absence an unmet clinical need and proposing that the surgical surveillance model be adapted.[6]
The transfer is an argument, not a finding, and the mechanisms are not the same. Surgery alters where and how nutrients are absorbed; these drugs reduce how much arrives. The overlap is the restriction, and the review literature is explicit that the inference from one to the other remains to be tested. What the surgical route actually involves is set out in the surgery comparison, and the people who have had both are covered in the post-surgical article.
The case reports are the sharp end, and they are case reports
Two published cases show what the severe version looks like when it arrives. A 65-year-old woman with type 2 diabetes presented with profound weakness, cranial nerve deficits, difficulty swallowing and a 120-pound weight loss after months of semaglutide with persistent nausea and vomiting. Serum thiamine, folate and vitamin B12 were all below normal; electromyography showed a severe sensory-predominant polyneuropathy; supplementation produced improvement in cognition and strength within five days and partial functional recovery.[7]
In the second, a 39-year-old man on semaglutide after bariatric revision surgery developed severe lower-limb weakness over three months that worsened as his dose rose to 1.7 mg weekly. Thiamine and vitamin E were deficient, imaging found no structural cause, and he improved over a month on supplementation and rehabilitation after the drug was stopped.[8]
Both are single patients, both involved an unusual degree of gastrointestinal intolerance or a stacked second weight-loss intervention, and neither establishes a frequency. What they establish is that the failure mode is real, that it is neurological rather than vague, and that prolonged vomiting is the common thread — the same pathway described in the vomiting article.
What this leaves
The solid statements are narrow. Intake falls by roughly a quarter to two-fifths in randomized trials. In those same trials, an objective marker of nutritional risk was found in 2.90% of treated participants against 1.77% on placebo, while investigators recorded malnutrition in 0.12%. The population starting these drugs already carries modestly worse diet quality and micronutrient adequacy than the population that does not. And the specific nutrient list in circulation is drawn from observational reports, mechanistic reasoning and the bariatric-surgery literature rather than from measurement in this class.
Nothing here is advice, and no one should read a nutrient list off this page and start buying things. Which of these measurements is worth making in a particular person depends on their baseline, their symptoms and their other conditions, which is a clinical judgment. It is also worth noting that a structured nutrition layer was present in both arms of the trials that included one, so a seller charging a premium for coaching is charging for the comparator — and most of that market dispenses compounded preparations that are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before dispensing, which is to say they were not in any of the trials above either.