What Are the GLIM Criteria for Malnutrition?

The GLIM criteria are a standardized framework for diagnosing malnutrition in adults, built around five clinical indicators split into two categories: three phenotypic criteria (what the body looks like) and two etiologic criteria (what is driving the problem). A diagnosis requires at least one criterion from each category. Published in 2019 by the Global Leadership Initiative on Malnutrition, the framework was the first attempt to give clinicians worldwide a shared language for identifying and grading malnutrition, replacing a patchwork of regional tools that often disagreed with one another.

Where the GLIM Criteria Came From

Before GLIM, hospitals and nutrition societies in different parts of the world used different screening and diagnostic tools. A patient could be classified as malnourished in one country and adequately nourished in another, depending on which assessment was used. In January 2016, several of the largest global clinical nutrition societies convened the Global Leadership Initiative on Malnutrition to fix this. Representatives from the European Society for Clinical Nutrition and Metabolism (ESPEN), the American Society for Parenteral and Enteral Nutrition (ASPEN), the Parenteral and Enteral Nutrition Society of Asia (PENSA), and the Latin American Federation of Parenteral and Enteral Nutrition (FELANPE) formed a core working committee, supported by a broader group with additional global diversity.¹ The group met in person at successive congresses and communicated between meetings by phone and email, building consensus through iterative discussion rather than a single vote.²

The result, published in 2019, was not meant to replace every existing screening tool. Instead, it proposed a two-step process: first, use whatever validated screening tool your institution already has in place to flag patients who might be at nutritional risk. Then, for those flagged patients, apply the GLIM criteria to confirm or rule out a malnutrition diagnosis and grade its severity.

The Five Criteria

The GLIM framework groups its criteria into phenotypic (observable changes in the body) and etiologic (causes or drivers of nutritional decline). A malnutrition diagnosis requires at least one phenotypic criterion and one etiologic criterion to be present simultaneously.³

Phenotypic Criteria

These describe what has happened to the patient’s body:

Weight loss: Unintentional loss of body weight over a defined period. Moderate malnutrition corresponds to a loss of 5–10% within the past six months or 10–20% beyond six months; severe malnutrition corresponds to greater losses.
Low BMI: A body mass index below certain thresholds that vary with age. For patients under 70, a BMI below 20 suggests moderate malnutrition, and below 18.5 suggests severe malnutrition. For those 70 and older, the thresholds shift upward slightly (below 22 for moderate, below 20 for severe).
Reduced muscle mass: Detectable loss of skeletal muscle, assessed by body composition methods or clinical examination.

Of the three, reduced muscle mass has proven the most challenging to assess in everyday practice. Weight loss and BMI require only a scale and a height measurement. Muscle mass, by contrast, can be estimated through body composition tools like bioelectrical impedance analysis (BIA), CT scans, or dual-energy X-ray absorptiometry (DXA), but many clinical settings lack routine access to these.⁴ When body composition technology is not available, the GLIM group endorses anthropometric measures like calf circumference or mid-upper arm circumference, as well as physical examination by a trained clinician.⁵

Etiologic Criteria

These describe why the nutritional decline is happening:

Reduced food intake or assimilation: The patient is eating less than normal, or their body is not properly digesting or absorbing what they eat. This could stem from poor appetite, nausea, difficulty swallowing, gastrointestinal symptoms, or conditions that impair absorption.
Inflammation or disease burden: The patient has an acute or chronic illness, infection, or injury that drives inflammatory processes, which in turn increase metabolic demands and break down muscle tissue.

The etiologic criteria serve a purpose beyond diagnosis. The GLIM group recommended that they be used to guide treatment planning and predict outcomes, because the underlying cause of malnutrition shapes what interventions are likely to help.³ A patient whose malnutrition is driven primarily by inflammation from cancer, for instance, requires a different care approach than someone who simply cannot eat enough after oral surgery.

Severity Grading

Once a diagnosis is made, GLIM grades it as either Stage 1 (moderate) or Stage 2 (severe) based on the phenotypic criteria. The severity grade is determined by whichever phenotypic criterion meets the more severe threshold. For example, a patient who has lost 12% of body weight in the last six months meets the severe weight-loss threshold regardless of whether their BMI is still in a moderate range. The etiologic criteria do not directly determine the severity grade, but they inform the treatment strategy and help classify the malnutrition into etiology-based diagnostic categories: chronic disease with inflammation, chronic disease without inflammation, acute disease or injury, and starvation-related malnutrition.

The Albumin Misconception

For decades, low serum albumin was treated as a marker of malnutrition. Many clinicians and hospital systems still reflexively check albumin levels when evaluating a patient’s nutritional status. The GLIM framework explicitly moves away from this practice. Albumin is a negative acute-phase reactant, meaning it drops sharply during inflammation, and strong consensus now holds that albumin lacks validity for diagnosing malnutrition in the presence of inflammatory conditions.⁶ A patient with pneumonia or a hip fracture will almost always have low albumin whether they are malnourished or not. Treating albumin as a malnutrition marker in those situations leads to overdiagnosis and muddies clinical decision-making.

That does not mean albumin is useless. The GLIM guidance acknowledges its value as a predictor of poor outcomes and as an indicator of inflammation itself. If you see a low albumin level, it should prompt you to look for inflammatory disease rather than to conclude malnutrition. When the contribution of inflammation to a patient’s condition is uncertain, the GLIM group recommends measuring C-reactive protein (CRP) to help fulfill the inflammation criterion.⁷ In acutely ill older patients, one study found that a CRP threshold of roughly 3.0 mg/dL appeared to be a reasonable cutoff indicating acute inflammation associated with reduced food intake.⁸ Importantly, though, the GLIM criteria allow clinicians to meet the inflammation criterion based on the clinical condition alone, without any lab test, when the disease is known to involve inflammatory activity.

How the Criteria Perform in Practice

A diagnostic framework is only as useful as its ability to predict real outcomes. Validation studies have tested GLIM criteria against older established tools and against hard clinical endpoints like mortality and length of stay. In hospitalized patients, one prospective cohort study found that a GLIM-based malnutrition diagnosis roughly doubled the odds of prolonged hospitalization and was associated with about five times the risk of dying during the hospital stay, along with nearly four times the risk of death within six months.⁹

When compared head-to-head with the Subjective Global Assessment (SGA), a widely used older tool, agreement between the two is generally good. A multicenter cross-sectional study in hospitalized patients found that GLIM criteria, when pooled across various criterion combinations, achieved about 61% sensitivity and 90% specificity relative to SGA.¹⁰ The relatively modest sensitivity means GLIM misses some patients that SGA would catch, largely because the muscle-mass criterion is often dropped or inconsistently measured in practice. On the flip side, the high specificity means that when GLIM does flag someone, the diagnosis is usually correct. In chronic heart failure patients specifically, a study found that GLIM and SGA showed good agreement (Cohen’s kappa of 0.8), and GLIM had a slightly better ability to predict adverse outcomes.¹¹

Cancer, Older Adults, and Other High-Risk Populations

Malnutrition is devastatingly common in cancer patients, and GLIM-defined malnutrition in this population carries serious prognostic weight. A meta-analysis found that cancer patients diagnosed as malnourished by GLIM criteria had about 75% higher all-cause mortality overall, with hazard ratios of roughly 1.44 for moderate malnutrition and 1.79 for severe malnutrition.¹² That dose-response pattern, where worse malnutrition grades predict worse survival, is exactly the kind of gradient you want to see in a diagnostic framework. It suggests the severity staging captures something real rather than simply being a label.

In older adults, a systematic review and meta-analysis found that GLIM criteria performed well for identifying malnutrition, with pooled sensitivity of about 79% and specificity of about 88%. GLIM-defined malnutrition in older patients was associated with higher in-hospital and post-discharge mortality, worse functional status at discharge, longer hospital stays, more complications, and higher healthcare costs.¹³ One notable gap: no validation studies were found in nursing homes or community-dwelling older adults, meaning the evidence is strongest for hospitalized and emergency department populations.

In acutely ill older adults admitted through the emergency ward, one study classified about half of patients as malnourished by GLIM criteria, with almost a third in the severe category. Severe malnutrition in that group was associated with more than four times the risk of in-hospital death.¹⁴

When BMI Hides the Problem

One of the more counterintuitive findings in recent malnutrition research is that patients with elevated BMI can still be malnourished by GLIM criteria. In older adults with high BMI, a study found that about 30% met GLIM-defined malnutrition. Strikingly, more than half of those patients were not flagged as at risk by the Mini Nutritional Assessment-Short Form (MNA-SF), a commonly used screening tool.¹⁵ This happens because many screening tools lean heavily on weight loss and low BMI, which may not apply to someone who is overweight but losing muscle mass and experiencing inflammation. In patients with sarcopenic obesity, where excess fat coexists with significant muscle loss, GLIM-defined malnutrition was even more common. The clinical implication is important: a higher body weight does not protect against malnutrition, and screening tools that rely too heavily on low BMI will miss these patients.

Using the Criteria in Low-Resource Settings

The muscle-mass criterion is the biggest practical barrier to GLIM adoption globally, because DXA machines, CT scanners, and even BIA devices are not available in many hospitals and clinics, particularly in low- and middle-income countries. Researchers have tested simple surrogate measures to see if they can stand in for formal body-composition analysis. In one validity study, substituting mid-upper arm circumference (MUAC) for the standard muscle mass index within the GLIM phenotypic criteria produced near-perfect agreement with the original criteria, with sensitivity of 100% and specificity above 90%. Calf circumference (CC) performed almost as well.¹⁶ Neither surrogate produced any false negatives, which is exactly the property you want in a screening-adjacent context: you catch everyone who truly has low muscle mass, even if a few extra patients are flagged. These findings suggest that a tape measure can do most of the work of an expensive body-composition device when the goal is GLIM diagnosis, a meaningful advance for resource-constrained health systems.

How the Dietary Intake Criterion Is Being Refined

The etiologic criterion for reduced food intake or assimilation was initially left somewhat open-ended, which made it hard for clinicians to apply consistently. How much of a decrease counts? Over what time frame? Should gastrointestinal symptoms like nausea or early satiety count even if caloric intake is not formally quantified? A modified Delphi study was conducted to build consensus on these questions, aiming to provide more specific guidance for identifying reduced dietary intake in clinical practice.¹⁷ This work reflects a broader pattern with the GLIM criteria: the original 2019 consensus established the framework, and subsequent working groups have been filling in the operational details criterion by criterion.

In a 2025 update, the GLIM group revisited and refined several elements of the framework, addressing areas where the original criteria were too vague for reliable use across different clinician skill levels and practice settings.¹⁸ The initiative has effectively become an ongoing project rather than a one-time consensus statement.

Coding, Reimbursement, and Institutional Adoption

A diagnostic framework only changes practice if health systems actually adopt it, and that adoption often hinges on whether the diagnosis translates into reimbursement. In many hospital payment systems, documenting a comorbidity like malnutrition can increase the reimbursement a hospital receives for a given admission, because the patient’s expected cost of care rises when malnutrition is present.¹⁹ For GLIM to become the default diagnostic tool, its criteria need to be recognized within the coding systems (like ICD-10) that hospitals use to document diagnoses and bill payers.

Currently, that integration is incomplete. Institutional variability in how malnutrition is screened, diagnosed, and documented means that the same patient could be coded differently depending on which hospital they end up in. A narrative review of GLIM implementation in gastrointestinal oncology highlighted this as a systemic challenge, noting that policy reforms recognizing GLIM-based malnutrition in coding systems could strengthen institutional incentives for adoption.²⁰ Without that policy link, there is a risk that GLIM remains an academic exercise at institutions where nutrition teams champion it while being ignored at facilities that lack dedicated dietitians or nutrition support teams.

What the Criteria Do Not Cover

The GLIM criteria were designed for adults. Pediatric malnutrition involves different growth dynamics, different reference standards, and different clinical considerations, and the GLIM group explicitly did not extend the framework to children or adolescents. Separate pediatric malnutrition definitions exist, and any attempt to apply GLIM thresholds to a growing child would be inappropriate.

The framework also does not prescribe a single screening tool. GLIM assumes that some form of validated nutrition screening has already taken place before the diagnostic criteria are applied. Which screening tool gets used, whether that is the Malnutrition Screening Tool, the Nutritional Risk Screening 2002, the MNA-SF, or something else, is left to the institution. This is a deliberate design choice: the GLIM group recognized that hospitals already have screening workflows in place and that mandating a switch would slow adoption. The downside is that different screening tools have different sensitivities, so the group of patients who make it to the GLIM assessment step varies depending on which screen was used upstream. The case of older adults with elevated BMI, where the MNA-SF missed more than half of those who met GLIM malnutrition criteria, illustrates how much the choice of screening tool matters.

Muscle-mass assessment using body composition tools also produces different values depending on the device and the population-specific cutoffs applied. A study in colorectal cancer patients found that the malnutrition rate varied depending on whether CT-based skeletal muscle index or BIA-based appendicular skeletal muscle mass index was used, with different sex-specific thresholds for each method.²¹ Validated ethnic- and sex-specific cutoffs are recommended when available, but in many populations they simply do not exist yet. This is an area of active research, and clinicians working with the criteria should expect the cutoff values to evolve as more population-level data become available.