A meta-analysis of 19 randomised trials behind semaglutide, tirzepatide and liraglutide reports a gap between what those trials recorded and what their own bloodwork showed. Investigators logged malnutrition as an adverse event in 0.12% of participants. Laboratory screening in the same trials found abnormally low lymphocyte counts in 2.90% of people taking active drug, against 1.77% on placebo. The authors argue that routine adverse-event reporting understates nutritional risk on these drugs, and propose baseline and periodic lab checks. It was published on 6 September in Obesity Science & Practice.
What the pooled trials measured
The review pooled 19 randomised trials drawn from the SURMOUNT (tirzepatide), STEP (semaglutide), SCALE (liraglutide) and OASIS (oral semaglutide) programmes, in adults with obesity. It followed PRISMA 2020, rated bias with Cochrane RoB2 and certainty with GRADE, and pooled continuous outcomes as mean differences in a random-effects model.
Daily energy intake fell by 24.00% to 39.20% across the drug classes, with model-estimated deficits reaching 1,200 kcal a day. Tirzepatide 15 mg was associated with a mean fat-free mass reduction of 1.60 kg. The abstract describes that as 2.80% of body weight, and those two figures do not reconcile: 1.60 kg is 2.80% of a 57 kg person, far below the entry weights in these obesity programmes, so one of the two numbers must rest on a different denominator. Pancreatic lipase rose by a mean of 31% in the SCALE programme.
The headline finding is the screening gap. A total lymphocyte count below 910 per microlitre turned up in 2.90% of active-therapy participants and 1.77% of placebo participants — an absolute difference of 1.13 percentage points, or roughly one additional person for every 89 treated. That is about 24 times the 0.12% rate investigators recorded as malnutrition.
What it changes, and what it does not
This does not overturn what earlier work found about muscle. Coverage of body composition on GLP-1 drugs has shown patients losing substantial weight over a year while grip strength held. The argument here is narrower and different: that the screening used in trials was not built to catch nutritional deterioration, so the recorded rate cannot be read as the real one.
The authors propose a tiered stepped-care algorithm — baseline albumin and total lymphocyte count, repeat testing at weeks 12, 24 and 52, and defined thresholds for intervention, with particular attention to adults aged 65 and over. That is a proposal from four researchers in the Department of Biochemistry and Biotechnology at Kwame Nkrumah University of Science and Technology in Ghana. No professional society has adopted it, and it should not be read as a standard of care. If it concerns you, it is a conversation for the clinician managing your treatment, not a reason to start buying supplements marketed at GLP-1 patients.
What is not yet known. A low lymphocyte count is not a diagnosis of malnutrition. It falls with infection, with corticosteroids and with acute illness generally, and this paper reports no hard outcomes — fractures, infections, hospital admissions — to anchor the marker to actual harm. The energy-intake figures are model-estimated rather than measured from food records. The 1.13-point gap rests on trials that were not designed to detect malnutrition in the first place, which is the authors’ own point but also a limit on how precisely it can be quantified. The 19 pooled trials were primarily funded by Eli Lilly and Novo Nordisk; the authors state this synthesis received no corporate support. And the fat-free-mass discrepancy above needs the full text to resolve.
Sources: Ampofo E, Apprey C, Amoako M, Turkson FD. “A Systematic Review and Meta-Analysis of Malnutrition and Metabolic Failure in High-Potency Incretin Therapy.” Obesity Science & Practice 2026;12(5):e70188, published online 6 September 2026. PubMed 42707648
