Diagnosing undernutrition children and adults: new French criteria. Why, for what and for whom? A joint statement of the French National Authority for Health and French Federation of Nutrition

Abstract

The objective was to establish new diagnostic criteria for undernutrition for the French population, concordant for children aged <18 years and adults aged <70 years, easy to use by health professionals and applicable whatever the situation (in and outpatients). A multi-disciplinary working and a reading group were involved. The procedure was divided into four phases: (1) systematic review and synthesis of the literature; (2) writing of the initial version of the guidelines; (3) reading and (4) finalisation. The literature search included international guidelines, meta-analyses, systematic reviews and randomised control trials from January 2007 to 31 July 2018. A two-step approach was selected: diagnosing undernutrition and then grading its severity. For diagnosis at least one phenotypic criterion associated with at least one aetiologic criterion were required for both children and adults. Phenotypic criteria for children were weight loss, Body Mass Index (BMI) < International Obesity Task Force curve 18·5, weight stagnation, reduction of muscle mass/function; for adults: weight loss, BMI < 18·5 and reduction of muscle mass/function. Aetiological criteria for children and adults were reduction in dietary intake, reduced absorption and hypercatabolism. Phenotypic metrics were used in both children and adults for grading severity (moderate or severe). These new French recommendations integrate the proposals of recent international recommendations combining aetiologic with phenotypic criteria, but for the first time, they are concordant for children and adults. The WHO threshold of 18·5 for BMI was kept as phenotypic criteria because epidemiological data show an increased mortality for that threshold.

Undernutrition can be defined as ‘the state of an organism in nutritional imbalance’; this imbalance is characterised by a negative energy and/or protein balance. The imbalance inherent in undernutrition leads to deleterious effects on tissues and/or the entire body^(1–4), with measurable changes in body function and/or composition associated with a worsening of the prognosis of the underlying disease^(2,4) . Within the concept of undernutrition, most authors have long recognised two major forms: the form without oedema has been classically called marasmus and the form with oedema has been called kwashiorkor^(5,6) . As summarised by Waterlow⁽⁷⁾, according to classical theory, kwashiorkor results from a protein deficiency with relatively adequate energy intake, whereas marasmus is caused by a global energy and protein deficiency. Note, however, that Gopalan et al. ⁽⁸⁾ proposed that the difference between marasmus and kwashiorkor could be explained by the child’s ability (marasmus) or not (kwashiorkor) to adapt. Actually, as underlined by Briend in his report⁽⁹⁾, the true pathophysiology of kwashiorkor remains incompletely elucidated. In practice, undernutrition with oedema appears to be more severe than the form without oedema because of possible complications, particularly digestive disorders and infections⁽⁸⁾; adults may also be affected⁽⁸⁾. Thus, the term proteolytic undernutrition may be more appropriate today. More recently, the terms ‘cachexia’, ‘sarcopenia’ and ‘fragility’ have appeared. Cachexia is characterised by an involuntary loss of muscle mass in a context of excessive cytokine production as in cancer and heart failure^(10–12) associated with a worsening prognosis. Sarcopenia initially defined by loss of skeletal muscle mass and is currently characterised by loss of muscle mass associated with functional deterioration. Frailty is a concept that combines vulnerability and poor adaptability as well as low energy and protein reserves^(13,14) .

In 2019, a group of experts in a consortium called the Global Leadership Initiative on Undernutrition (GLIM) published globally applicable recommendations for adults⁽¹⁵⁾. The main novelty was the combination of so-called phenotypic criteria with so-called aetiological criteria. The experts selected five criteria: involuntary weight loss, decrease in Body Mass Index (BMI), reduction in muscle mass, decrease in food intake or absorption and presence of disease/inflammation. The proposed diagnostic criteria combine at least one phenotypic and at least one aetiologic criterion. These recommendations were a very significant step forward mainly because they incorporate the notion of aetiological criteria such as an underlying disease associated with phenotypic criteria which were generally the only ones taken into consideration up to now. They also incorporate the importance of quantifying muscle mass. With regard to paediatric undernutrition, the most widely used classification system was that proposed by Waterlow⁽⁷⁾. He suggested that acute undernutrition (wasting) be defined independently of age using weight as a percentage of weight for height calculated from the 50th percentile of the Boston standard⁽¹⁶⁾ and chronic undernutrition (stunting) be defined by height for age again based on the Boston standard. Acute undernutrition and chronic undernutrition were divided into four stages. Then, other definitions have been proposed^(17–20). In 2013, the Pediatric Undernutrition Definitions Working Group⁽²¹⁾ indicated that five areas should be considered: anthropometric parameters, growth, chronicity of undernutrition, aetiology and pathogenesis and developmental/functional outcomes. They classified undernutrition as acute or chronic with a threshold of 3 months duration for the latter, with or without a disease-related threshold for aetiology, the presence or absence of an inflammatory state and pathogenic mechanisms leading to suboptimal nutrient intake/absorption. They recommended that anthropometric measures be expressed as z-scores, as proposed by the WHO⁽¹⁷⁾. Their consensus statement for 2015⁽²²⁾ recommended the use of the following indicators: food/nutrient intake, assessment of energy and protein requirements, growth parameters, speed of weight gain, mid-upper arm circumference, grip strength, indirect measures replacing traditional anthropometric measures and documentation of Tanner stage. The experts proposed to use z-scores for the criteria of weight for height/length, BMI for age, length/height for age or mid-upper arm circumference is indicated. They define mild undernutrition as a z-score between –1 and –1·9 sd, moderate undernutrition as a z-score between –2 and –2·9 sd and severe undernutrition as a z-score ≤ 3 sd for each indicator. This American group proposed that indicators of undernutrition, i.e. weight for height, BMI or height for age, be used with the WHO growth standards for children from birth to 2 years of age and the CDC growth charts for children aged 2–20 years⁽¹⁸⁾.

In summary, in recent years, expert committees have proposed new recommendations for the diagnosis of undernutrition in adults and children. So why propose French recommendations now? For the following main reasons: (a) the previous French recommendations were obsolete and dated from 2003, with several inconsistencies; (b) we disagreed with the choice of the 20 kg/m² threshold proposed by GLIM as a phenotypic criterion for undernutrition in adults under 70 years of age; (c) we wanted the diagnostic criteria for undernutrition in children to be consistent with those for adults and (d) in all cases, we wanted the diagnosis criteria for adults and children to be easy to use in clinical practice.

Methods

Good practice guidelines are defined in the health field as ‘methodically developed proposals to help the practitioner and the patient find the most appropriate care in given clinical circumstances’. These guidelines were then developed according to Haute Autorité de Santé (HAS) standards. The complete methodology is available in the HAS methodological guide available on its website (https://www.has-sante.fr/jcms/c_431294/fr/recommandations-pour-la-pratique-clinique-rpc).

The clinical practice guidelines method is a rigorous method based on:

• participation of professionals;

• transparency, with the provision of a critical analysis of the literature; the essential points of the debates and decisions made by the members of the working group; notes and comments from the members of the reading group and the list of all the participants in the different groups.

• independence linked to the status of the HAS as an independent public scientific authority (Law of 13 August 2004 on health insurance, Title II, Chapter I bis, Article L. 161–37);

• independence of the groups from one another; the working and reading groups each have a specific role which they perform independently of one another;

• financial independence; public financing in the context of the HAS GPG;

• management of the interests declared by the experts of the working group, according to the procedure described in the HAS ‘Guide on declaration of interests and management of conflicts of interest’.

In the current study, the clinical practice guidelines method involved two groups: a working group and a reading group. Two project managers were in charge of coordinating all the work of the working group with the HAS project manager. The working group was composed of twenty-one professionals: four nutritionists, three paediatricians, three general practitioners, one biologist, one oncologist, one gastroenterologists, two geriatricians, one dietitian, one pharmacist, one intensive care anaesthetist, one nurse and a HAS project leader. The reading group was composed of thirty-four professionals concerned with the subject. Like the working group, this group was multidisciplinary and multi-professional. The procedure of the method is divided into four phases: (1) systematic review and synthesis of the literature; (2) writing of the initial version of the guidelines; (3) reading and (4) finalisation.

Systematic review and literature synthesis phase

The drafting of the evidence report was preceded by a phase of document search and critical analysis of the literature. The reviewers, the project managers, HAS project leader and scientific librarian participated in the creation of the document search strategy. The document search was systematic, hierarchical and structured.

A strategy was designed to search MEDLINE (National Library of Medicine, USA), The Cochrane Library (Wiley Interscience, USA), Science Direct (Elsevier) and the HTA (International Network of Agencies for Health Technology Assessment) database using the following search terms (MeSH terms and equivalent free text terms):

For children: (1) First step: ‘Malnutrition/diagnosis’[Majr] OR ‘Severe Acute Malnutrition/diagnosis’[Majr] OR ‘Nutrition Assessment’[Majr] OR ‘Child Nutrition Disorders/diagnosis’[Majr] OR ‘Nutritional Status’[Majr] OR (Nutritional Or malnutrition Or malnourished OR undernutrition OR undernutrition OR undernourished OR under-nourished) [title] AND (Screen or screening or assessment or measuring OR evaluation Or diagnostic OR diagnosis OR detection Or criteria Or definition) [title] AND consensus OR guidance OR guide OR guidelines OR position paper OR recommendation* OR statement Field: Title; (2) Second step: ‘Malnutrition/diagnosis’[Majr] OR ‘Severe Acute Malnutrition/diagnosis’[Majr] OR ‘Nutrition Assessment’[Majr] OR ‘Child Nutrition Disorders/diagnosis’[Majr] OR ‘Nutritional Status’[Majr] OR (Nutritional Or malnutrition Or malnourished OR undernutrition OR undernutrition OR undernourished OR under-nourished) [title] AND (Screen or screening or assessment or measuring OR evaluation Or diagnostic OR diagnosis OR detection Or criteria Or definition) [title] AND ‘Meta-Analysis as Topic’[Mesh] OR ‘Meta-Analysis ’[Publication Type] OR ‘Review Literature as Topic’[Mesh] OR ‘Meta-Analysis’ OR ‘Systematic Review’ OR ‘Literature Review’ Or ‘Quantitative Review’ OR ‘Pooled Analysis’ [Title/Abstract]

For adults: (1) First step: ‘Malnutrition’[MeSH Major Topic] OR ‘Severe Acute Malnutrition’[MeSH Major Topic] OR ‘Nutrition Assessment’[MeSH Major Topic] OR ‘Nutritional Status’[MeSH Major Topic] OR ((‘Malnutrition’[Title] AND (‘Or’[All Fields] AND ‘malnourished’[Title])) OR ‘undernutrition’[Title] OR ‘undernutrition’[Title] OR ‘undernourished’[Title]) AND Guideline* or Consensus OR guidance OR recommend *[title]; (2) Second step: ‘Malnutrition’[MeSH Major Topic] OR ‘Severe Acute Malnutrition’[MeSH Major Topic] OR ‘Nutrition Assessment’[MeSH Major Topic] OR ‘Nutritional Status’[MeSH Major Topic] OR ((‘Malnutrition’[Title] AND (‘Or’[All Fields] AND ‘malnourished’[Title])) OR ‘undernutrition’[Title] OR ‘undernutrition’[Title] OR ‘undernourished’[Title]) AND ‘Meta-Analysis as Topic’[Mesh] OR ‘Meta-Analysis ’[Publication Type] OR ‘Review Literature as Topic’[Mesh] OR ‘Meta-Analysis’ OR ‘systematic Review’ OR ‘Literature review’ Or ‘Quantitative Review’ OR ‘pooled analysis’ [Title/Abstract].

The languages selected were English and French. The start date used was January 2007, and the last search date used was 31 July 2018. Several articles in the bibliography appeared to be published outside this period because (1) the search strategy has been updated regularly until the end of the project in 2019 and (2) the search strategy was supplemented by the bibliographic contribution of the experts of the working group and reading groups, and the references cited in the documents analysed. The literature search included international guidelines, meta-analyses, systematic reviews and randomised controlled trials.

The literature search was not limited to articles published and indexed in databases. For this, grey literature was found by consulting relevant sources. This search made it possible to initially identify the French and international guidelines and evidence reports created by governmental agencies, independent evaluation agencies and learned societies. French and international biomedical databases were queried. It was supplemented by the bibliographic contribution of the experts of the working group and reading groups, and the references cited in the documents analysed. The document search strategy appeared in the evidence report available on the HAS website: –https://www.has-sante.fr/upload/docs/application/pdf/2019–11/reco277_argumentaire_rbp_denutrition__cd_2019_11_13_v0.pdf. It describes the key words used as well as the types of documents searched in the databases, specifying the results obtained, and also states the sources used for searching grey literature.

A total of 1258 publications were identified. Three reviewers (JCD, FJ, EM) reviewed all titles and selected studies based on titles and/or abstracts. Studies that met the defined inclusion criteria were selected for article review. If it was not clear from the abstract whether a study met the inclusion criteria, the full article was reviewed.

The flow chart of literature screening is presented in Fig. 1. The reviewers conducted a critical analysis and synthesis of the selected literature in the form of an evidence report and proposed a list of recommendations based on the literature review conducted. The proposed recommendations based on the critical review of the literature by the reviewers were sent to the members of the working group 15 d before the first meeting.

Fig. 1 Flow chart of literature screening.

Drafting of the initial version of the guidelines

The members of the working group met for five one-day discussion sessions between April 2018 and January 2019, in order to create, on the basis of the evidence report and the proposed recommendations drafted by the reviewers, the initial version of the guidelines to be submitted to the reading group. The two project leaders of the working group and the HAS project leader shared the responsibility for leading and facilitating the working group meetings. During the working group meetings, the evidence reports and proposed recommendations were discussed on the basis of existing data and practices. The majority of the recommendations were based on the agreement of the experts in the working group and were then referred to as ‘expert consensus’. The members of the working group were appointed by HAS on the proposal of the parties concerned by the topic: national professional specialty councils, the Board of general medicine, professional organisations and institutions. For this guideline project, the working group included twenty-one French professionals from different specialties: nutrition, paediatrics, general medicine, oncology, gastroenterology, geriatrics, pharmacy, biology, nursing, and dietetics.

The composition of the experts group guaranteed a balanced representation of:

• main healthcare professions in accordance with the project outline;

• methods of practice (public, university-based or not and self-employed);

• different currents of opinion or schools of thought;

• the geographical origins of the experts.

In general, the healthcare professionals of the group must have a good knowledge of professional practice in the field corresponding to the topic of the study and must be able to judge the relevance of the published studies and different clinical situations evaluated. The working group must bring together professionals concerned by the topic under consideration.

The absence of a rating does not mean that the guidelines are not relevant and useful, but should nevertheless prompt further study. They should apply to the majority of cases, sometimes with an adjustment on a case-by-case basis.

Reading phase

In March 2019, the HAS project leader sent the evidence report, the initial version of the guidelines and the questionnaire to the members of the reading group, upon which each member gave an individual opinion electronically (using the GRaAL computer tool, available on the HAS website www.has-sante.fr). The questionnaire includes a discrete numerical scale, ranked from 1 to 9, and a free text area for each recommendation made. It allows each member of the reading group to judge the form and content of the initial version of the guidelines, as well as the acceptability, applicability and readability of each recommendation. The score ranges from 1 (Strong Disagree) to 9 (Strongly Agree). In order to improve the final text, any rating below 5 should be accompanied by a commentary.

Finalisation

In May 2019, the final version of the guidelines was drafted at the last meeting of the working group. After analysis and discussion of the reading group’s notes and comments, the initial recommendations were modified according to the following rules: a) when the Reading Group confirmed the appropriateness of the recommendation (≥90 % of the Reading Group’s answers within the range (5–9)), the recommendation is retained and the relevant comments are taken into account to improve the form; b) when the reading group is more broadly undecided or disagrees with the initial recommendation (<90 % of the reading group’s responses within the range (5–9)), the working group discusses the appropriateness of the comments and, if necessary, modifies the recommendation. If discussions in the meeting reveal divergent views, a vote in the working group session should confirm the withdrawal or final wording of the amended recommendation. Ten recommendations had then been modified. Finally, 191 publications were discussed in the Evidence Report. This last step involved the working group and the HAS validation bodies. The final versions of the Evidence Report, the recommendations and its synthesis were published in November 2019 on the HAS website (https://www.has-sante.fr/upload/docs/application/pdf/2019-11/reco277_argumentaire_rbp_denutrition__cd_2019_11_13_v0.pdf).

Results

General recommendations, criteria for diagnosing, assessing the severity and monitoring undernutrition in children (<18 years)

The general recommendations are reported in Table 1. Undernutrition should be screened at each consultation, recorded in any document and can only be diagnosed in the presence of at least 1 phenotypic criterion and 1 aetiological criterion. This diagnosis is compulsory first before judging its severity. The phenotypic criterion is based exclusively on non-biological criteria.

Table 1. General recommendations for diagnosing undernutrition in children (<18 years)

The phenotypic and aetiological criteria (one of each is sufficient) are reported in Table 2. Phenotypic criteria are weight loss, BMI < 18·5 (IOTF curve), weight located two corridors below the usual corridor (weight curve) and reduction in muscle mass/function. Aetiologic criteria are reduced dietary intake, decrease in food assimilation or absorption and hypercatabolism with or without inflammation.

Table 2. Phenotypic and etiological criteria for diagnosing undernutrition in children (<18 years)

* The International Obesity Task Force (IOTF)’s curves are those recommended by the National Nutrition Health Plan for monitoring children’s BMI. However, the IOTF does not propose BMI curves before the age of 2. The IOTF’s curves were extended by the ‘AFPA –- CRESS/Inserm - CompuGroup Medical 2018’ curves over this age range. Curves available on the website: https://cress-umr1153.fr/index.php/courbes-carnet-de-sante

† Usual corridor = child’s usual or reference weight growth corridor for specific diseases (Down syndrome, myopathy, etc.).

‡ There are no validated methods for assessing muscle function in children. Various methods have been described, such as the distance covered in 7 min, but none of them has been validated as a criterion of undernutrition in children. For muscle mass, there is no consensus on the reference value of brachial perimeter or brachial circumference. Validation studies are therefore necessary.

^§ Reduced absorption should be considered in cases of chronic diarrhoea or extensive small bowel resection or in cases of biological stigma of maldigestion such as reduced fat-soluble vitamins plasma concentrations.

The assessment of body composition can be done by different methods depending on local possibilities. Absorptiometry is the reference method, but other techniques such as impedancemetry can be used. However, these are 20-year-old works that cannot currently be used as references and justify studies in children in France. Measurement of the brachial perimeter, recognised by the WHO, is another simple method for assessing muscle mass. However, the threshold defining undernutrition remains debated as it varies according to the populations studied: <115 mm between 6 and 60 months for the WHO or <133 mm between 6 and 60 months in the Asian population, which is known to have a lower corpulence. This does not allow recommendations to be made. The 6-minute walk test is probably the simplest and most reproducible, but no standard has yet been established in children and requires further study.

Albumin was not used as a diagnosis criterion, because it is not a reliable marker of undernutrition in children due to the variations related to other conditions, such as inflammatory syndrome or enteropathy. In kwashiorkor, hypoalbuminaemia is a marker of hypoproteinaemia and of the presence of oedemas. In marasmus, albumin may be normal.

Once diagnosis of undernutrition has been done, its severity must be assessed.

The assessment criteria of the severity of undernutrition are reported in Table 3. Criteria for moderate undernutrition (one is sufficient) are BMI > 17 and <18·5 (IOTF curve), ≥5 % weight loss in 1 month or ≥10 % in 6 months or ≥10 % as compared with the usual weight before the start of the disease, weight located two and three corridors below the usual corridor. Criteria for severe undernutrition (one is sufficient) are BMI ≤ 17 (IOTF curve), >10 % weight loss in 1 month or >15 % in 6 months compared with the usual weight before the start of the disease, weight located at least three corridors (representing three standard deviations) below the usual corridor, loss of height of at least one corridor compared with the usual height.

Table 3. Assessment of severity of undernutrition in children*

* When a single criterion for severe undernutrition is observed simultaneously with one or more criteria for moderate undernutrition, a diagnosis of severe undernutrition is recommended.

IOTF, International Obesity Task Force.

The monitoring of the changes in nutritional status of children is reported in Table 4. If undernourished, adapting nutritional care according to the level of severity, especially preventing refeeding syndrome; in outpatient care, systematically re-evaluating the nutritional status in the month following the last assessment; during hospitalisation, reassessing the nutritional status at least once a week is recommended. If not undernourished but with a disease (i.e., at risk of malnutrition) in outpatient care, reassessing the nutritional status at each consultation during hospitalisation on admission.

Table 4. Monitoring the change in nutritional status of children

General recommendations, criteria for diagnosing, assessing the severity and monitoring undernutrition in adults (≥18 to <70 years)

The general recommendations for diagnosing undernutrition in adults (≥18–<70 years) are reported in Table 5. Undernutrition should be screened at each consultation, recorded in any document and can only be diagnosed in the presence of at least 1 phenotypic criterion and 1 aetiological criterion. This diagnosis is compulsory first before judging its severity. The phenotypic criterion is based exclusively on non-biological criteria.

Table 5. General recommendations for diagnosing undernutrition in adults (≥18 to <70 years)

The phenotypic and aetiological criteria (one of each is sufficient) are reported in Table 6. Phenotypic criteria are weight loss, BMI < 18·5, reduction in muscle mass/function. Aetiologic criteria are reduced dietary intake, decrease in food assimilation or absorption and hypercatabolism with or without inflammation.

Table 6. Phenotypic and aetiological criteria for diagnosing undernutrition in adults (≥18 to <70 years)

* Reduced absorption should be considered in cases of chronic diarrhea or extensive small bowel resection or in cases of biological stigma of maldigestion such as reduced fat-soluble vitamins plasma concentrations.

The proposed methods and thresholds for the quantification of the reduction in muscle mass and/or function according to the most recently available data are reported in Table 7: grip strength (dynamometer), walking speed, muscle surface area index in L3 in cm²/m² (CT-scan, MRI-scan), muscle mass index in kg/m² (impedance measurement), non-fat mass index (impedance measurement) in kg/m² and appendix muscle mass (DEXA) in kg/m².

Table 7. Proposed methods and thresholds for the quantification of the reduction in muscle mass and/or function according to the most recently available data. References are in parenthesis

Once diagnosis of undernutrition has been done, its severity must be assessed.

The assessment of the severity of undernutrition is reported in Table 8. Criteria for moderate undernutrition (one is sufficient) are BMI > 17 and <18·5, ≥5 % weight loss in 1 month or ≥10 % in 6 months or ≥10 % as compared with the usual weight before the start of the disease, albuminaemia >30 g/l and <35 g/l. Criteria for severe undernutrition (one is sufficient) are BMI ≤ 17, >10 % weight loss in 1 month or >15 % in 6 months compared with the usual weight before the start of the disease and albuminaemia ≤ 30 g/l.

Table 8. Assessment of the severity of undernutrition in adults aged 18–69 years*

* When a single criterion for severe undernutrition is observed simultaneously with one or more criteria for moderate undernutrition, a diagnosis of severe undernutrition is recommended.

The monitoring of the changes of nutritional status is reported in Table 9. If undernourished, adapt the nutritional care to the level of severity, preventing the refeeding syndrome. In outpatient care, assessing nutritional state within 3 months of the last assessment. If hospitalisation, reassessing at least once a week. In follow-up consultations after hospitalisation, systematically reassessing. If not undernourished, but with a disease (i.e., at risk of malnutrition) in ambulatory care, reassessing at each consultation. If inpatient care, reassessing on admission.

Table 9 Monitoring of the change of nutritional status

Discussion

These new French recommendations for diagnosis of undernutrition of both children and adults aged less 70 years were undertaken because previous adult French recommendations (published in 2003)⁽²³⁾ needed to be revised for many objective reasons, among which: (a) there were no paediatric French recommendations; (b) we wanted a consistency between paediatric and adults; c) we disagreed with the choice of a BMI threshold of twenty proposed by the GLIM consortium even we recognised that their recommendations had several qualities. We will focus the discussion about the choice of the threshold of BMI, the significance of weight loss, the importance of the decrease in food intake, the true significance of plasma albumin concentrations and the consistency between children and adults’ recommendations. The recommendations for adults >70 years of age first published in 2003⁽²³⁾ were revised in 2007⁽²⁴⁾ and are currently under new revision. For the sake of brevity, the GLIM criteria that we have retained will not be discussed below.

Why did we choose a threshold of 18·5 for BMI as a phenotypic criterion?

Before discussing the choice of a threshold for BMI, it is useful to briefly recall its history (Tables 2 and 6). Weight/height² (W/H²) was proposed by Quetelet in 1832⁽²⁵⁾, a Belgian mathematician, astronomer and statistician⁽²⁶⁾, not to define nutritional state but to define the characteristics of ‘normal man’. ‘Now, if we compare fully developed and regularly built individuals with each other, in order to know the relations that may exist between weight and height, we will find that the weights in developed individuals of different heights are about the same as the squares of the heights’^(25,27) . W/H² was called Quetelet’s indice. Keys et al. renamed it in 1972 as the BMI⁽²⁸⁾. The authors showed, in twelve cohorts from five countries that BMI was related to fat mass, but no more than half of the total variance in fat mass was accounted for by the regression of fat mass on BMI. Durnin et al. ⁽²⁹⁾ showed that among 6000 healthy and fit young men (5000) and women (1000) in the British Army, the percentage of fat was 16·6–21·1 and 27·2–29·8, respectively. In fact, BMI includes both fat and lean tissue. In terms of body size, women’s bodies have a higher percentage of fat and a lower muscle mass than men’s⁽³⁰⁾ and women’s urinary creatinine size index, a biomarker of muscle mass, is lower than men’s⁽³¹⁾. During famine, in the absence of disease, both adipose tissue and lean tissue (muscle) are used as fuel, but the proportion of lean tissue lost depends on the amount of fat stored⁽³²⁾: the more adipose tissue mass, the lower the loss of lean tissue⁽³³⁾. The preferential loss of lean tissue is the determining factor in an individual’s survival at low body weight. During the course of a disease, not only does muscle mass begin to decrease but there is also an increase in muscle fatigability⁽³⁴⁾. Muscle strength and endurance can be assessed simply by measuring the strength and durability of the grip⁽³⁵⁾.

The lower limit of 18·5 kg/m² was defined in the report of a working group of the International Food Energy Consultative Group sponsored by the United Nations University and the Subcommittee on Nutrition of the United Nations System, published in 1988 by James et al. ⁽³⁶⁾. Based on data for normal males and females, they concluded that the upper limit for the diagnosis of diagnosis of chronic energy deficiency (CED) should be 18·5 kg/m² because a BMI > 18·5 kg/m² was consistent with good health in both male soldiers and normal females on the basis of data from Durnin et al. ⁽²⁹⁾ for the healthy population and data from healthy adults in the Third World. The WHO report⁽³⁷⁾⁾ used the average BMI for British Army men aged 25–40 years of 24·6 kg/m² and of 22·7 kg/m² for women aged 25–35 years⁽²⁹⁾ and took – 2 sd as the threshold for the lower limit of an acceptable range, so that the weighted average lower limit was 18·5 for men and 17·6 for women. In most groups of adults living in low-income countries identified by Eveleth & Tanner⁽³⁸⁾, the mean BMI was between 19 and 21 kg/m². Based on these data, James et al. ⁽³⁶⁾ finally proposed three thresholds at 16, 17, and 18·5, such that a BMI ≥ 18·5 kg/m² was considered normal (non-CED), 17·0–18·4 kg/m² was classified as CED grade I, 16·0–16·9 kg/m² was classified as CED grade II, and <16·0 kg/m² as CED grade III. This classification was approved by a meeting of the IDECG in 1992⁽³⁹⁾, and then by the WHO Expert Committee on Physical Status in 1993, but CED was renamed ‘thinness’⁽⁴⁰⁾. The report states ‘a BMI below 16 is known to be associated with a markedly increased risk of ill-health, poor physical performance, lethargy and even death, so this cut-off point has validity as an extreme limit. Moreover, BMI below 17 kg/m² has been linked with a clear-cut increase in illness in adults studied in three continents and is therefore a further reasonable value to choose as a cut-off point for moderate risk. The proposal of a single cut-off point of 18·5 kg/m² for specified mild deficiency in both sexes has less experimental support, but seems a reasonable value to use pending further comprehensive studies’. The current undernutrition threshold proposed by the WHO is 18·5 (http://apps.who.int/bmi/index.jsp?introPage=intro_3.html). This threshold is also recommended in France for adults under 70 years of age by the National Nutrition and Health Program (in French: Programme National Nutrition Santé) (http://www.sante.gouv.fr/IMG/pdf/brochure_denutrition.pdf), by the decree of November 9, 2009 for the management of enteral nutrition at home (http://textes.droit.org/JORF/2010/02/24/0046/0033/) and for coverage by French national health insurance in adults under 70 years old. (https://www.legifrance.gouv.fr/affichTexte.do?cidTexte=JORFTEXT000021394117&categorieLien=id).

In addition, with regard to children, in 2007, the IOTF established thresholds for defining ‘thinness’ in six countries (Brazil, the UK, Hong Kong, the Netherlands, Singapore and the USA), with a total of 97 876 boys and 94 851 girls aged 0–25 years measured in the years 1968–1993⁽¹⁹⁾. These were equivalent thresholds for underweight, based on the WHO definition of underweight in adults, which corresponds to a BMI of 18·5 kg/m² (level 1 underweight), 17 kg/m² (level 2 underweight) or 16 kg/m² (level 3 underweight) at age 18 years. Percentile curves were plotted to cross the threshold of BMI 17 to 18 years of age. The resulting curves were averaged to provide age- and sex-specific threshold points from 2 to 18 years of age. Similar threshold points were derived based on BMI 16 and 18·5 kg/m² at 18 years of age, together providing definitions of thinness levels 1, 2 and 3 for children and adolescents, consistent with WHO definitions for adults. For children under 2 years of age, WHO standards for growth curves from birth to 5 years of age were established in 2006⁽⁴¹⁾ and are recommended for use in all countries⁽⁴²⁾. However, a systematic study⁽⁴³⁾ showed that these growth curves were imperfectly calibrated with the growth of contemporary children in many countries, including France. In 2019, new growth curves for French children were established from 238 102 children (1 458 468 height measurements and 1 690 340 weight measurements), using a ‘big-data’ approach⁽⁴⁴⁾, and from these data, updated BMI curves were established for girls⁽⁴⁵⁾ and boys⁽⁴⁶⁾ respectively.

Mortality is increased when BMI is <18·5 kg/m²

Mortality has been repeatedly shown to increase significantly in people with a BMI < 18·5 kg/m². Calle et al. ⁽⁴⁷⁾ studied the relationship between BMI and all-cause deaths in a prospective American cohort of 457 785 men and 588 369 women followed for 14 years. Among white men and women who had never smoked and were disease free at study entry, mortality increased by 26 % and 36 %, respectively. Flegal et al. ⁽⁴⁸⁾ showed from the three national health and nutrition surveys (NHANES I, 1971–1975; NHANES II, 1976–1980; NHANES III, 1988–1994, to 2000) (571 042 person-years of follow-up), an increase in mortality of 38 % and 130 % among participants aged 25–59 and 60–69 years, respectively. In a subsequent study using NHANES I, II, III data, the same authors⁽⁴⁹⁾ found a 3·6-fold increase in mortality from non-cancer and non-cardiovascular (CV) causes. Jee et al. ⁽⁵⁰⁾ examined the association between BMI and mortality in a cohort of 1 213 829 Koreans (12-year follow-up, 82 372 deaths from all causes and 48 731 deaths due to specific diseases). After adjusting for age, smoking, alcohol consumption, exercise, fasting blood glucose, systolic blood pressure and serum cholesterol, total mortality increased by 51 % in men and 25 % in women for a BMI < 18·5 kg/m². He et al, ⁽⁵¹⁾ used data from the National Survey of Hypertension in China, including a representative sample of the general Chinese population ≥15 years (1 239 191 person-years of follow-up, mean follow-up time 8·3 years, 20 033 deaths). Mortality increased by 47 % among participants with a BMI < 18·5 kg/m² (RR = 1·47; 95 % CI: 1·42, 1·53). In a prospective American cohort (Cancer Prevention Study II, 891 572 white and 38 119 black men and women, 28-year follow-up), Patel et al. ⁽⁵²⁾ found an RR for all-cause deaths associated with low BMI (15–18·5) of 1·25 (95 % CI: 1·08, 1·45) in white men who had never smoked without prevalent disease, but not significantly in black men (RR = 0·78, 95 % CI: 0·28, 2·13). Among white women, the RR was 1·20 (95 % CI: 1·14, 1·26), and among black women it was 1·38 (95 % CI: 1·03, 1·85). The Global BMI Mortality Collaboration⁽²⁶⁾, in 2016, conducted a meta-analysis of 239 prospective studies on four continents (10 625 411 participants, median follow-up 13·7 years). The main objective was to limit the analysis to healthy, non-smoking individuals and to exclude the first 5-year follow-up in order to limit confounding effects and reverse causality. Underweight (BMI 15·0 to <18·5 kg/m²) was associated with a 47 % increase in mortality (RR = 1·47; 95 % CI: 1·39, 1·55). Afzal et al. ⁽⁵³⁾ analyzed data from three Danish cohorts (1976–1978, 1991–1994, 2003–2013) followed until the end of 2014. A BMI < 18·5 kg/m² was associated with a 63 % (RR = 1·63; 95 % CI: 1·43, 1·86), 78 % (RR = 1·78; 95 % CI: 1·47, 2·15) and 68 % (RR = 1·68; 95 % CI: 1·35, 2·08) increase in all-cause mortality for the 3 cohorts, respectively.

Thus, all these data in Westerners (white and black) and Asians indicate that a BMI < 18·5 kg/m² is associated with excess mortality in both men and women. The higher threshold of 20 kg/m² proposed by GLIM experts was chosen on the basis of the average increase in BMI of the entire American adult population over the last few years. The GLIM report⁽¹⁵⁾, which is intended for the world community, states that ‘the experience of the current USA population shows that people are often overweight or obese and that they would have to lose a lot of weight before being assigned a low BMI. However, there is no reference cited to support the claim that mortality is increased for a BMI < 20 kg/m² in people under 70 years of age. This other sentence in the GLIM recommendations ‘However, further research is needed to obtain consensus baseline BMI data for Asian populations in clinical settings’ indicates that if 20 is proposed, it cannot be for the global nutrition community. We have cited several studies showing clearly that a BMI < 18·5 kg/m² is associated with mortality in the same way in Westerners as in Asians. For all these reasons, we have found that a threshold of 18·5 as a phenotypic criterion is the most appropriate.

Since BMI is strongly related to FM and does not necessarily correlate with muscle mass, an assessment of muscle mass or function requires further measurement. Two muscle-specific metabolites were evaluated as potential measures of body muscle: creatinine and 3-methylhistidine. With respect to creatinine, Heymsfield et al. ⁽⁵⁴⁾ concluded that urinary creatinine had inter- and intra-individual variability and that a defined value for creatinine equivalence with a range of kg muscle mass per g urinary creatinine from 17 to 22 was missing. As for urinary 3-methylhistidine, which mainly reflects protein renewal and not muscle mass, it requires a totally meat-free diet for 3 days and sedentary activity; the coefficient of variation of collection is 5 to 10 % and it requires a stable state of protein renewal, which is not obtained in case of famine or catabolic diseases. The methods/techniques currently proposed to assess muscle mass/function are: grip strength (dynamometer), walking speed, muscle surface area index in L3 (CT-scan, MRI-scan), muscle mass index (impedance measurement) and appendix muscle mass (DEXA). Reference values have been proposed^(55–57), and research is underway to better delineate these references.

Why is weight loss important for diagnosis of undernutrition?

Andres et al. ⁽⁵⁸⁾, in 1993, reviewed thirteen studies and concluded that even slight or moderate long-term weight loss was generally associated with a high mortality rate (Tables 2 and 6). As reviewed by Wong⁽⁵⁹⁾, several other studies showed that involuntary weight loss was associated with a mortality rate of 16–38 %^(60–65). Allison et al. ⁽⁶⁶⁾, in 1999, compared the effect of weight loss in two cohorts: the Tecumseh Community Health Study (1890 subjects; 321 deaths within 16 years of follow-up) and the Framingham Cardiac Study (2731 subjects; 507 deaths within 8 years of follow-up). They found that each sd of weight loss (4·6 kg in Tecumseh, 6·7 kg in Framingham) increased mortality by 29 % and 39 %, respectively. In contrast, each sd of fat loss (10·0 mm in Tecumseh, 4·8 mm in Framingham) reduced the risk of mortality by 15 % and 17 %, respectively. Bullock et al. ⁽⁶⁷⁾ conducted a meta-analysis on markers of undernutrition and clinical outcomes in elderly cancer patients. One study found that 5 % weight loss in 3 months was associated with early post-operative death within 3 months; 2 studies found an association between weight loss and mortality, where weight loss between 5 and 10 %, >10 %, >3 kg or unknown weight loss was associated with 1 year mortality. Weight loss in the last 6 months was also associated with mortality. DeWys et al. ⁽⁶⁸⁾ found that weight loss was also a predictor of reduced survival by approximately 50 % in 3047 cancer patients, depending on tumor type, grade and stage. A 5 % decrease in weight was sufficient to decrease survival. Similarly, in patients with respiratory disease, weight loss is associated with increased mortality and disability⁽⁶⁹⁾. In patients with small cell lung cancer, among 25 factors, weight loss was the fourth most associated with decreased survival⁽⁷⁰⁾.

Losing weight is not always harmful. Losing fat is sometimes desirable, but losing weight becomes harmful when weight loss is mainly FFM. Since FFM contains water (about 80 %), and assuming that it is composed mainly of proteins, 1 g (dry weight) of protein actually weighs 5 g in normally hydrated tissue. Thus, 1 g of FFM represents about 1 kcal, while 1 g of fat mass (which does not contain water) contains 9 kcal. Therefore, 1800 kcal corresponds to 200 g of fat or 1·8 kg of muscle. In real life, when a patient loses weight, she or he consumes both fat and non-fat mass, but in varying proportions depending on the situation. For a given energy deficit, the faster the rate of weight loss, the greater the proportion of muscle consumed. Thus, the speed of weight loss is recognised as a phenotypic criterion of undernutrition by many consensuses, including the GLIM. This makes it possible to diagnose undernutrition in overweight and obese patients. The thresholds we have chosen are different from those of the GLIM (GLIM has chosen a 5 % weight loss in 6 months, whereas we have chosen 10 % for the same period of time), but close to those previously published such as the Subjective Global Assessment (SGA)⁽⁷¹⁾, the Malnutrition Universal Screening Tool (MUST)⁽⁷²⁾ or the Nutritional Risk Screening 2002 (NRS-2002)⁽⁷³⁾. For a person weighing 80 kg, for example, losing 4 kg (5 %) in 6 months corresponds to an energy deficit of 36 000 kcal (200 kcal per day) if she or he consumes 4 kg of fat. Theoretically, such a weight loss can occur without necessarily affecting muscle mass. This is much less likely when the patient loses 10 % of their body weight during the same period. In other words, we have chosen a threshold that is less sensitive but more specific to a decrease in muscle mass.

In many consensuses/recommendations, weight loss must be unintentional to be considered as a diagnosis criterion for undernutrition. However, this remains poorly supported by scientific evidence. Indeed, even if loss of weight is intentional (e.g., after bariatric surgery), loss of muscle mass remains harmful. We understand that inducing undernutrition by trying to treat obesity is a problem for many colleagues, but while weight loss is sometimes desired, the loss of lean mass is never voluntary and that of muscle mass not desirable. This is why, based on pathophysiology rather than on the treatment of obesity, we decided that any weight loss corresponding to the thresholds we had chosen should be considered as a phenotypical criterion of undernutrition, even in the case of voluntary weight loss.

Of course, depending on the clinical situation and the expected benefit of voluntary weight loss, permissive undernutrition can be tolerated. This is particularly the case in young adults. But it seemed important to us to stress that any voluntary weight loss must be the subject of a risk benefit analysis and that priority should be given to slow and continuous weight loss associated with physical activity.

Why is decrease in food intake important too?

In adults, the dominant cause of decreased body weight is a decrease in food consumption, caused either by the unavailability of sufficient food to meet energy requirements or by anorexia or any other cause limiting the patient’s ability to eat (Tables 2 and 6). For example, in elderly patients with cancer, a recent meta-analysis found that decreased food consumption was associated with mortality (OR 2·1; P < 0·00001)⁽⁶⁷⁾. Assessment of dietary intake is therefore of major importance in the evaluation of nutritional status. In patients with small cell lung cancer, 97 % experienced a loss of appetite⁽⁷⁴⁾. Among twenty-five factors, loss of appetite was the sixth most common factor associated with decreased survival⁽⁶⁷⁾.

What about albumin?

In people who have a low caloric intake, such as patients with anorexia nervosa or hunger strikers, with no somatic disease, serum albumin levels remain normal even at very low BMIs⁽⁷⁵⁾ (Table 8). In addition, for many years, it has been shown that plasma albumin decreases in inflammatory states such as injuries when the acute phase reaction is activated. This decrease is the result of both decreased synthesis and increased catabolism⁽⁷⁶⁾. Low plasma albumin levels have been shown to be a prognostic factor for increased mortality or poor outcomes in many diseases such as end-stage renal failure, cancer and surgery. Mortality and poor prognosis are inversely related to plasma albumin levels^(77–82). Thus, albumin should not be considered as a marker of undernutrition but as a marker of severity. Since albumin is a linear prognostic factor (the lower it is, the worse the prognosis), there is no clearly defined cut-off. We have retained the value of 30 g/l which corresponds to a curvature of the albumin/prognosis relationship.

We wanted the criteria to be as consistent as possible between children and adults and as easy as possible to use

In order to achieve this objective, it was necessary that an expert committee comprising specialists in adult and paediatric medicine formulate the recommendations for adults and children at the same time. In addition, we wanted the criteria to be easy to use and understand by non-nutrition health professionals, so that we would not have to propose too many criteria. Finally, we felt it was very necessary to take into account the new and very relevant approach used by both the GLIM consensus and the Academy of Nutrition and Dietetics/ASPEN consensus statement. This very interesting approach mixes phenotypic and aetiological criteria and highlights the importance of assessing muscle mass/function. It is clear that the aim of the new French recommendations was not to be orthogonal to the two recommendations cited, which were intended for the world nutrition community, but to be applicable to the French community, which only had at its disposal dated recommendations that no longer appeared relevant for adults, and no recommendations for children. At the same time, we would also like to draw modest attention to some limitations we found in these adult and paediatric recommendations aiming to be used by the worldwide nutrition community.

Conclusion

These new French recommendations for the diagnosis of undernutrition for people under 70 years of age include several original features. They integrate the proposals of recent international recommendations combining aetiologic criteria with phenotypic criteria, but for the first time, they have been deliberately established to allow a concordance of adult and paediatric criteria. Regarding BMI curves in children, they propose to use both IOTF for children >2 years and new French BMI curves for children <2 years of age, so that the recommendations are applicable to children from 0 to 18 years of age. The WHO threshold of 18·5 for BMI proposed by the WHO was kept as phenotypic criteria because, among others things, epidemiological data in many populations including Asians showed an increased mortality for a BMI under that threshold. These recommendations will be revised every 3 to 5 years in order to take into account any new data likely to change the criteria and their applicability. Lastly, recommendations for people aged ≥70 years are ongoing and will be available in 2021.