Bioimpedance Scale or DEXA: How to Tell If You're Losing Fat or Muscle

A bioimpedance scale weighs you accurately, but it doesn’t measure fat or muscle: it calculates them with a formula from the electrical current that passes through your body. In studies that compare these scales with DEXA, the error for any one person can be several kilograms^[1,4]. DEXA (or DXA, a whole-body scan) is the clinical reference, but it doesn’t see muscle on its own either. If you take a GLP-1 medication and worry about losing muscle, the most useful signals are the scale’s trend under the same conditions, your waist and, above all, your strength.

Why muscle is a concern on these medications, and how to protect it, is covered in protein and muscle on GLP-1 medications. Here I focus on how to measure it.

What each method measures

The bioimpedance scale

The scale sends a weak current through you and measures the body’s resistance, which depends mostly on water. The National Institutes of Health (NIH) technology assessment conference on bioimpedance concluded that it gives a reliable estimate of total body water under most conditions^[6]. Fat and muscle are then calculated with an equation. The European consensus on sarcopenia, EWGSOP2, puts it plainly: bioimpedance doesn’t measure muscle mass directly, it estimates it, and the equation is calibrated against DEXA in a specific population^[17]. If your body isn’t like that population’s, the estimate drifts.

According to Frija-Masson, most smart scales measure foot to foot only; other devices, like some at pharmacies or gyms, add hand electrodes^[1].

The 2025 advisory on GLP-1 therapy from four US obesity, nutrition and lifestyle medicine societies notes one limit: people with a pacemaker, an implantable cardioverter defibrillator or another electronic medical implant cannot use a bioimpedance machine^[16].

DEXA

It’s a scan with X-rays at two energies, done lying down; a 2021 review estimates 10 to 15 minutes per person^[12]. According to the International Society for Clinical Densitometry (ISCD), the report should include, among other data, bone density, total mass, total lean mass, total fat mass and body fat percentage. It may add visceral fat or the lean mass of the arms and legs^[11]. The dose from the most widely used machines is about 4 to 5 microsieverts, less than one day of natural background radiation in the United Kingdom (5 to 8 microsieverts), according to the same review^[12].

It isn’t perfect either. EWGSOP2 warns that machines from different brands don’t give consistent results and that hydration affects DEXA too^[17]. The 2021 review notes that the scanning area is about 195 cm (6 ft 5 in) long: people who are very tall or very broad may not fit entirely^[12]. And against a four-compartment reference method, Minderico (2008) found that DEXA overestimated the fat lost by 48 women who lost an average of 3.3 kg (7.3 lb). The authors concluded that none of the clinical methods they tested was accurate enough to track those changes and that they shouldn’t be used interchangeably^[5].

Skinfolds and tape measure

According to the 2021 review, skinfolds are the method least affected by meals or hydration, but they depend heavily on the skill of the person measuring. And because the equations that turn skinfolds into a body fat percentage disagree, the authors prefer comparing the sum of millimeters^[12].

A tape measure is the cheapest option and has published risk cut-offs (more below). The mirror, photos in the same light or how a pair of pants fits have no studies behind them, but they’re free and don’t depend on any formula.

Is a bioimpedance scale accurate? What the studies say

Frija-Masson concluded that smart scales aren’t accurate for body composition and shouldn’t replace DEXA^[1]. Bosy-Westphal, that foot-to-foot scales work for studying groups and that, for an individual, a device with hand and foot electrodes is preferable^[2]. Siedler, that some models (especially eight-electrode and certain foot-to-foot devices) may be useful for tracking change over time, where they performed best^[3].

Two more caveats:

• Repeatable isn’t accurate. In Siedler, measuring twice in a row gave almost the same number (precision error of 0 to 0.49 points)^[3]. A scale that repeats a number isn’t necessarily giving the right one.
• With obesity or dehydration, more caution. The ESPEN guidelines, from the European society for clinical nutrition, don’t recommend routine bioimpedance at extremes of BMI or with abnormal hydration. They consider tracking change with it possible at a BMI of 16 to 34 without abnormal hydration, and even then with caution^[7].

What moves the reading from day to day

The NIH conference listed what affects the measurement: body position, hydration, food and drink, air and skin temperature, and recent physical activity. It concluded that reliable bioimpedance requires controlling these variables^[6]. The specific studies:

• Meals. In 18 adults measured 18 times over 24 hours, impedance dropped for 2 to 4 hours after each meal, and the effect added up over the day. According to the abstract, calculated body fat percentage varied by 8.8% in women and 9.9% in men between the highest and lowest readings of the day, without specifying whether these are body fat points or a relative change (Slinde, 2001)^[8].
• Eating or drinking right before. In 55 adults on a foot-to-foot scale, a meal or an electrolyte drink shifted body fat percentage by about one point over the next two hours (Androutsos, 2015). The authors considered the effect small^[9].
• Menstrual cycle. In 25 women measured daily for one cycle, resistance and estimated fat-free mass changed between phases, though body fat percentage didn’t. Small weight changes from water, linked to salt intake, explained part of the error (Gleichauf, 1989)^[10].

On a GLP-1 medication there’s one more factor: if you’ve had vomiting or diarrhea, your hydration has changed, and so has the reading.

How to use the scale so it’s worth something

These rules follow from the above:

• Same conditions every time: when you get up, after using the bathroom and before eating or drinking.
• Never right after a workout, or after a day of vomiting or diarrhea.
• If you menstruate, compare readings from the same phase of your cycle.
• Look at the trend over weeks, not one day’s number.
• Don’t mix devices. According to EWGSOP2, muscle estimates vary with the brand of device^[17]; your home scale and the gym’s aren’t comparable.

What the semaglutide and tirzepatide trials measured with DEXA

The two main trials measured body composition with DEXA in a subgroup of participants^[13,14]. The STEP 1 kilograms are those reported in the 2025 advisory^[16].

In STEP 1, lean mass fell in kilograms, but its share of the body rose 3.0 points because much more fat was lost^[13]. The Neeland review (2024) notes that figures vary across GLP-1 studies: in some, lean mass was 40% to 60% of the weight lost; in others, 15% or less^[15].

Lean mass isn’t the same as muscle. Neeland explains that besides muscle it includes organs, fluids and the water inside fat tissue^[15]. The 2025 advisory estimates that muscle is about half of lean mass: in STEP 1, around 20% of the weight lost^[16].

When a DEXA scan is worth it and how to ask for one

The 2025 advisory recommends assessing strength, physical function and, where possible, muscle mass (by bioimpedance, DEXA or other methods) when treatment starts. For follow-up, it suggests DEXA every year or every two years. It also warns that many imaging sites with DEXA don’t run the body composition analysis^[16]. EWGSOP2, for its part, advises DEXA in clinical practice to confirm sarcopenia (loss of muscle mass and strength) once strength has tested low^[17].

If you think it might help you, bring it up at your next appointment. Three useful questions:

• Do my age, strength or other conditions justify measuring body composition?
• If so, can it be a whole-body DEXA with body composition analysis, not just a bone scan?
• If I repeat it, can it be on the same machine? Results from different brands aren’t comparable^[17].

You can write them down in the doctor-visit checklist.

Waist size: the cheap measure with published cut-offs

The WHO expert consultation lists these cut-offs for risk of metabolic complications, based mostly on studies in people of European origin^[18]:

The International Diabetes Federation (IDF) uses cut-offs by ethnic group. For people of Central and South American origin, it provisionally applies the South Asian values until more specific data are available. For people of European origin, 94 and 80 cm. It also notes that in the USA the higher values of 102 cm (40 in) for men and 88 cm (34.6 in) for women are likely to remain in clinical use^[19].

The WHO describes how to measure: standing, feet close together, wearing little clothing, midway between the lowest rib and the top of the hip bone. Keep the tape parallel to the floor and snug without squeezing, and read it at the end of a normal breath out. Measure twice and, if the two readings are within 1 cm (0.4 in), take the average^[18]. For tracking yourself, what matters is always measuring at the same spot. In what BMI measures and what it misses I explain why waist size complements BMI.

Signs you may be losing muscle, and what protects it

EWGSOP2 treats low strength as the main feature of sarcopenia, ahead of mass. These are the signs it lists as reasons to look for it: falls, feeling weak, walking more slowly, difficulty getting up from a chair, or losing weight or muscle. And these are its strength cut-offs, designed mainly for older adults^[17]:

• Rising from a chair five times without using your arms: more than 15 seconds indicates low leg strength.
• Grip strength, measured with a dynamometer: under 27 kg (about 60 lb) in men and 16 kg (about 35 lb) in women.

The 2025 advisory adds that these tests work better in older adults and may be less able to detect change in younger people^[16]. If you train, logging what you lift, or how many times you can stand up from a chair in 30 seconds (a variant of the test), gives you a signal no scale measures. If you notice any of the signs, talk to your doctor.

Two things together protect muscle. Enough protein: targets of 1.2 to 1.6 g per kg a day (about 0.55 to 0.73 g per lb) have been proposed during active weight loss, the 2025 advisory notes^[16], and you can work out your number with the protein calculator. And strength training: the 2025 advisory proposes at least three sessions a week and warns that more protein without strength training is likely not enough^[16]. The details are in protein and muscle on GLP-1 medications.

Frequently asked questions

Is a bioimpedance scale accurate?

For weight, yes. For one person’s fat and muscle, not very: in studies, the error reached several kilograms or more than 10 body fat points depending on the model^[1,2,3]. It’s more useful for seeing a trend, always measuring under the same conditions.

How much muscle do you lose on Ozempic?

On semaglutide 2.4 mg weekly, the Wegovy dose, lean mass was 38% of the weight lost in the STEP 1 subgroup (5.3 of 13.6 kg, or 11.7 of 30 lb)^[13,16]. On tirzepatide, about 25% in SURMOUNT-1^[14]. Not all of it is muscle: the 2025 advisory estimates muscle is about half^[16].

My scale says I’m losing muscle. Is it true?

The scale can’t confirm it. Some lean mass is lost with weight loss, but the scale’s muscle estimate is off by several kilograms and shifts with food and hydration^[1,8,9]. Watch your strength and the trend over several weeks.

Where can I get a DEXA scan?

Hospitals and imaging centers with a DEXA machine do them; ask your doctor whether it’s appropriate in your case. Ask for a whole-body scan with body composition analysis: according to the 2025 advisory, many sites with DEXA don’t offer it^[16].

Which is better, the scale or a tape measure?

They measure different things. Your waist doesn’t tell you how much muscle you have, but it’s cheap, easy to repeat and has risk cut-offs published by the WHO and the IDF^[18,19]. The scale estimates fat and muscle, with a large error. Together, and with your strength, they give a fuller picture.