Three GLP-1 changes your scale can't show. A body check-in shows two.

By Matt Cole — research host

TL;DR: Three studies landed in eleven days, describing one blind spot from three angles: hair, lean tissue, and bone all change during GLP-1 treatment, and a scale reports none of them. Two of the three you can rate at home in ten seconds a week. The third has no home answer at all.

Key Takeaways

• A scale returns one number — it can't separate fat from lean tissue, and doesn't register hair or bone. Three papers from July 12–22, 2026 each describe something it summarizes away.
• Hair: real, and small. A BMJ study found a 37% higher alopecia rate than with SGLT-2 inhibitors — 6.91 vs 5.04 per 1,000 person-years — concentrated in the non-scarring type, where follicles stay intact.
• Muscle: about 30% of the weight lost is lean mass, per a meta-analysis of nine placebo-controlled trials.
• Bone: the 2026 evidence points both ways — which is why it's a scan-and-clinician question, not a home-tracking one.
• The behaviour that answers all three is a weekly body check-in: pick the loudest signal, rate it 1 to 5, done. Six weeks of that beats "I think it's been worse lately."

---

Weigh yourself and you get one number. It's a good number. It's also a summary — and a summary works by discarding the terms it summarized. This month three groups published on three tissues in eleven days. Read together, they aren't three warnings.

Signal one: hair

Using University of Pennsylvania Health System records, researchers ran a target trial emulation — a design that mimics a randomized trial with real-world data — on adults with type 2 diabetes newly starting a GLP-1. Published in The BMJ on July 22, it found a 37% higher alopecia rate than with SGLT-2 inhibitors (6.91 vs 5.04 per 1,000 person-years) and 68% higher than with DPP-4 inhibitors (6.53 vs 3.89) (Tang et al., BMJ, 2026).

Two details missed the headlines. Absolute rates run roughly 3 to 9 per 1,000 people a year — a 37% increase on a small base is still a small base. And the association was specific to non-scarring alopecia, where the follicle stays intact, which the authors note leaves potential for regrowth. That's about a category of hair loss, not a promise about anyone's hair.

A separate analysis landed that day: nference matched 11,046 people on tirzepatide against the same number on semaglutide and found new-onset alopecia in 4.24% versus 3.33% — 5.44% versus 3.63% among women. The gap held regardless of weight lost or dose, and alopecia was likelier in people with existing thyroid or endocrine conditions, pointing at baseline hormonal factors rather than rapid loss alone (Reuters, July 22, 2026). It's a preprint, not yet peer-reviewed — a safety observation, not a steer toward one medication. That's a prescriber conversation. The fuller picture — timing, mechanism, and what the evidence does and does not support — is in what the 2026 research found about GLP-1 hair loss.

Signal two: muscle

The cleanest number comes from a meta-analysis of nine placebo-controlled trials (659 participants) using DXA: 6.9 kg of total weight loss, 1.9 kg of it lean mass — about 30% of the total (Beavers et al., Obesity, 2024).

A narrative review published July 12 in Obesity extends the question to older adults, where muscle and bone already decline with age. Its authors found six trials showing significant losses in lean soft tissue and three showing losses in bone indices — while stating plainly that the clinical significance for function and fracture risk remains unclear (Saavedra et al., Obesity, 2026).

That "unclear" isn't a hedge — it's the state of the field, and why the move is noticing rather than panicking. If you've never put a number on your protein target, our protein calculator gives you one to bring to that conversation, and what the body-composition trials found goes through the numbers in full.

Signal three: bone — and why it sits differently

Here the evidence pulls both ways. The July review flagged bone-index losses in three trials. But a June study in the same journal, using 18,062 propensity-matched adults 50 and over without diabetes, found GLP-1 users had lower osteoporosis and fracture risks than users of other anti-obesity medications (Wu et al., Obesity, 2026).

Two credible papers, same journal, same season, opposite signs — that's what an unsettled question looks like. It settles a practical point. Bone density is measured by a DXA scan, not by paying attention. Any app implying you can track it at home is selling you something. Bone belongs with your prescriber — the evidence on both sides is laid out in full there.

What the three have in common

Different tissues, different mechanisms, different certainty — one shared blind spot: each can be moving while the scale reads exactly what you hoped. It's the same gap behind a stall that isn't really a stall and what the regain research does and doesn't say — one number asked to carry a whole body. You don't close that by weighing more often. You close it with a second instrument.

The behaviour: a weekly Body Signals check-in

In Gila, that instrument is Body Signals, and it asks one question: What's your body telling you? You pick the loudest signal — Hair shedding and Muscle changes are both on the list, alongside the GI and mood signals — and rate it 1 to 5. Ten seconds a week, in English, Turkish, or Spanish.

Two of this month's three signals live there. The third doesn't, and we won't pretend otherwise.

The severity rating is what earns its keep. A yes/no toggle says something happened. A 1-to-5 trend says whether it's building, holding, or fading — without relying on your memory of six weeks ago, the least reliable instrument here. "Hair shedding, mostly 3s since week nine, ticked to 4 this month" is a sentence a clinician can work with. "I think it's been worse lately" isn't. That's the payoff: not a diagnosis, a better handoff. (We also wrote a comparison of GLP-1 tracking apps, Gila included.)

None of this is a reason to change anything about your medication — not the dose, not stopping. Titration, timing, supplements, and every version of "should I stay on this" belong with the person who prescribed it. What's yours to hold is whether you noticed, and whether you wrote it down.

The scale will keep giving you its one number. Keep it — just stop asking it questions it was never built to answer.

---

Try the check-in. Gila's pilot is open — Body Signals, food noise, and the habit side of all this. Join the pilot.

Or start with the research. One plain-language read on new GLP-1 studies, weekly. Subscribe to the newsletter.

---

Sources

• Tang H, Zhang B, Lu Y, et al. "Risk of hair loss associated with glucagon-like peptide-1 receptor agonists in adults with type 2 diabetes: target trial emulation." BMJ, July 22, 2026 — doi:10.1136/bmj-2026-100077 · publisher summary
• Lapid N. "Hair loss with GLP-1 drugs is rare, but real, studies find." Reuters, July 22, 2026 — reporting the nference preprint, not yet peer-reviewed — Reuters wire
• Beavers KM, Cortes TM, Foy CM, et al. "GLP1Ra-based therapies and DXA-acquired musculoskeletal health outcomes: a focused meta-analysis of placebo-controlled trials." Obesity, December 22, 2024 — doi:10.1002/oby.24172
• Saavedra JM, Chen AS, Porter Starr KN, Batsis JA. "The Effects of Incretin Mimetic Therapies on Muscle and Bone Health in Older Adults: A Narrative Review." Obesity, July 12, 2026 — doi:10.1002/oby.70203
• Wu JS, Lin LY, Sun JW, et al. "Musculoskeletal Outcomes of Glucagon-Like Peptide-1 Receptor Agonists Versus Other Antiobesity Agents in Nondiabetic Adults." Obesity, June 1, 2026 — doi:10.1002/oby.70218

Gila is a companion app, not a medical provider. Nothing here is medical advice. Questions about your medication, your dose, or a symptom you're noticing belong with your prescriber.