Risk of malnutrition (over and under-nutrition): validation of the JaNuS screening tool

lable at ScienceDirect

Clinical Nutrition xxx (2013) 1e8

Contents lists avai

Clinical Nutrition

journal homepage: ht tp: / /www.elsevier .com/locate/c lnu

Original article

Risk of malnutrition (over and under-nutrition): Validation of theJaNuS screening tool

Lorenzo M. Donini a,*, Laura Maria Ricciardi b, Barbara Neri a, Andrea Lenzi a,Giulio Marchesini b

aDepartment Experimental Medicine, Medical Physiopathology, Food Science and Endocrinology Section, Food Science and Human Nutrition Research Unit,Sapienza University of Rome, 00185 Roma, ItalybDepartment of Medicine and Surgery, Unit of Metabolic Diseases & Clinical Dietetics, “Alma Mater Studiorum” University of Bologna, Italy

a r t i c l e i n f o

Article history:Received 9 July 2013Accepted 5 December 2013

Keywords:MalnutritionOvernutritionUndernutritionScreening toolJANUS

* Corresponding author. Tel.: þ39 064991 0996,fax: þ39 06 4991 0699.

E-mail address: [email protected]

0261-5614/$ e see front matter � 2013 Elsevier Ltd ahttp://dx.doi.org/10.1016/j.clnu.2013.12.001

Please cite this article in press as: Donini LClinical Nutrition (2013), http://dx.doi.org/1

s u m m a r y

Background & aims: Malnutrition (over and under-nutrition) is highly prevalent in patients admitted tohospital and it is a well-known risk factor for increased morbidity and mortality. Nutritional problemsare often misdiagnosed, and especially the coexistence of over and undernutrition is not usuallyrecognized. We aimed to develop and validate a screening tool for the easy detection and reporting ofboth undernutrition and overnutrition, specifically identifying the clinical conditions where the twotypes of malnutrition coexist.Methods: The study consisted of three phases: 1) selection of an appropriate study population (esti-mation sample) and of the hospital admission parameters to identify overnutrition and undernutrition;2) combination of selected variables to create a screening tool to assess the nutritional risk in case ofundernutrition, overnutrition, or the copresence of both the conditions, to be used by non-specialisthealth care professionals; 3) validation of the screening tool in a different patient sample (validationsample).Results: Two groups of variables (12 for undernutrition, 7 for overnutrition) were identified in separatelogistic models for their correlation with the outcome variables. Both models showed high efficacy,sensitivity and specificity (overnutrition, 97.7%, 99.6%, 66.6%, respectively; undernutrition, 84.4%, 83.6%,84.8%). The logistic models were used to construct a two-faced test (named JaNuS e Just A NutritionalScreening) fitting into a two-dimension Cartesian coordinate graphic system. In the validation samplethe JaNuS test confirmed its predictive value. Internal consistency and testeretest analysis provide ev-idence for the reliability of the test.Conclusion: The study provides a screening tool for the assessment of the nutritional risk, based onparameters easy-to-use by health care personnel lacking nutritional competence and characterized byexcellent predictive validity. The test might be confidently applied in the clinical setting to determine theimportance of malnutrition (including the copresence of over and undernutrition) as a risk factor formorbidity and mortality.

� 2013 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

1. Introduction

The health status is intimately connected with the nutritionalstatus; the maintenance of health status corresponds to themaintenance of the structural and functional body entirety (bodycomposition and body function, respectively) through energy (en-ergy balance) and substance exchange (energy and non-energy

þ39 3385926464 (mobile);

(L.M. Donini).

nd European Society for Clinical N

M, et al., Risk of malnutrition0.1016/j.clnu.2013.12.001

nutrients) with the environment.1 Any alteration of this fragilebalance generates malnutrition, a term which encompasses un-dernutrition, overnutrition, or the copresence of both the condi-tions, as in sarcopenic obesity.2 Malnutrition is highly prevalent inpatients admitted to hospital and it is a well-known risk factor forincreased morbidity and mortality, specifically in frail populationsand in the elderly.3,4

Although studies continue to highlight the high incidence ofundernutrition in patients, the increasing incidence of obesity inthe general population suggests that an ever increasing number ofobese patients will be admitted to hospital. Several studies havealso revealed an increased risk of complications in obese

utrition and Metabolism. All rights reserved.

(over and under-nutrition): Validation of the JaNuS screening tool,

L.M. Donini et al. / Clinical Nutrition xxx (2013) 1e82

hospitalized patients that may be the result of inappropriatenutritional support in this group. This situation may be a conse-quence of inappropriate screening and/or nutritional assessment.5

Nutritional problems are often misdiagnosed, and especially thecopresence of over and undernutrition is not usually recognized.6

In fact the early assessment of malnutrition is pivotal in the clin-ical setting and specifically in subjects admitted to hospital. In in-hospital patients, undiagnosed malnutrition (both over andunder-nutrition) is expected to prolong hospital stay, the use ofhealth care services and worsening comorbidities, functionalimpairment and frailty, finally affecting quality of life andsurvival.7,8

A number of screening tests have been developed for theassessment of undernutrition (e.g., the Mini-Nutritional Assess-ment (MNA),9 the Malnutrition Universal Screening Tool (MUST),10

the Nutritional Risk Screening (NRS),11 the Subjective GlobalAssessment (SGA)12).

Early identification of undernutrition precludes any assessmentof risk in obese patients. The precision and reliability of anyscreening tool assessing risk of undernutrition, when applied toobese patients, is likely to show these patients as having anacceptable nutritional status.6 Therefore obese subjects, by defini-tion, do not seem to be at nutritional risk and clinical conditions,such as concomitant sarcopenia, unintentional weight loss andreduced food intake, are not correctly identified by the currentscreening tests.

Since screening tools aimed at identifying the copresence ofover and undernutrition are scarce,6 we aimed to develop andvalidate a screening tool for the easy detection and reporting ofboth undernutrition and overnutrition, specifically identifying theclinical conditions when the two types of malnutrition coexist.

2. Materials and methods

The study consisted of three phases:

1. Selection of the appropriate study population (estimationsample) and of the parameters to be evaluated at the admissionin order to identify:- Overnutrition, defined as a fat mass increased by 25% in menand by 35% inwomen, respectively, with or without metabolicor cardiovascular complications13;

- Undernutrition, defined as: 1) reduced triceps skinfold thick-ness: 2) reduced muscle arm circumference; 3) reduced levelsof transport plasma proteins; 4) low lymphocyte count;

2. Combination of the above mentioned variables to create ascreening tool to be used by non-specialist health care pro-fessionals, able to assess the nutritional risk in case of under-nutrition, overnutrition, or the copresence of both theconditions;

3. Validation of the screening tool in a different patient sample(validation sample).

In the first phase, participants were recruited among patientsadmitted to Food Science and HumanNutrition Research Unit of theSapienza University. The following admission data wereconsidered:

History: a) family history of disease, stroke, venous thrombo-embolism, dyslipidemia, type 2 diabetes; b) smoking habits (morethan 10 cigarettes per day).

Clinical status: a) presence of prediabetes/diabetes (IFG,impaired fasting glucose; IGT, impaired glucose tolerance; T2DM,type 2 diabetes)14; b) dyslipidemia (total cholesterol levels>200 mg/dL, triglyceride levels >150 mg/dL, HDL-cholesterol<45 mg/dL in men or <50 in women; LDL-cholesterol level

Please cite this article in press as: Donini LM, et al., Risk of malnutritionClinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.12.001

>130 mg/dL15; c) atherosclerosis; d) hypertension (including theuse and the number of anti-hypertensive agents); e) respiratorydiseases (COPD, chronic obstructive pulmonary disease, OSAS,Obstructive sleep apnea syndrome); f) osteoarticular disease; g)chronic renal failure; h) liver diseases;

Comorbidity level: it was assessed using the four-level Index ofCo-Existent Disease (ICED)16 (absence of symptoms; presence ofmild-to-moderate symptoms, responsive to treatment; presence ofsevere symptoms, incompletely responsive to treatment; severeend-stage disease, non-responsive to treatment). The presence ofpressure sores grade III and IV (full thickness skin defect includingmuscle, and destruction including bone and or joint structures,respectively) was also considered according to Shea’sclassification.17

Cognitive status: it was evaluated using the Short PortableMental Status Questionnaire (SPMSQ), a 10-item screening toolexploring cognitive skills. The total score ranges from 0 (intactmental functioning) to 10 (severe cognitive impairment).18

Autonomy level and functional parameters: Autonomy wasassessed using the Katz’s index of Independence in Activities ofDaily Living (ADLs) that counts the number of ADLs a person needshelp with (control of urination and bowelmovements, autonomy infeeding, bathing, dressing, toileting and autonomy in transfer).19

The forearm flexor muscle strength, expressed in kilograms, wastested using the Lafayette’s dynamometer, Mod. 78011. The test wasperformed using the dominant arm, except in hemiplegic patients,in whom the contralateral arm was considered.

Eating behavior: it was assessed by the SCOFF Questionnaireconsisting of 5 questions addressing the core features of anorexiaand bulimia nervosa (a score �2 indicates a risk of eatingdisorder).20

Antropometric measurements: a) Body weight, height and recentchanges in body weight; b) arm (AC), calf (CC) and waist circum-ference (WC); c) biceps (BSF), triceps (TSF), subscapular (SSSF) andsupra-iliac (SISF) skinfold thicknesses; d) fat mass. Anthropometricmeasures were obtained as described in the Standard Manual forAnthropometric Measures.21 Body weight was measured using aSECA scale (Hamburg, Germany) to the nearest 0.1 kg; height wasmeasured using a SECA stadiometer (Hamburg, Germany) to thenearest 0.5 cm. Arm, calf and waist circumferences were obtainedto the nearest 0.1 cm using a stretch-resistant tape; the measure-ment of skinfold thicknesses was performed using a Harpendenplicometer (British Indicators Ltd, St Albans, Herts, UK) to thenearest 0.2 mm. Body mass index (BMI ¼ body weight/heightsquared) and arm muscle circumference [AMC ¼ AC � (TSF*p)]were calculated; fat mass was estimated according to Durnin andWomersly equation22 and used to assess body density. Fat masspercentage was calculated from body density using Siri equation.23

Biochemical parameters: a) Lipid profile: total cholesterol levels,HDL-cholesterol, LDL-cholesterol, triglyceride levels; b) Glucosemetabolism: basal glucose and insulin levels, oral glucose tolerancetest considering both glucose and insulin response; c) Plasmaprotein profile: albumin, prealbumin and transferrin levels; d)Blood cell count (red and white cell count); e) Inflammatorymarkers: C-reactive protein (CRP) and mucoproteins.

Biochemical parameters were assayed using commercial kits.Blood samples were drawn from an antecubital vein after anovernight fast.

Mini-Nutritional Assessment (MNA)9: the individual items ofMNAwere considered, and specifically the anthropometric indexes(BMI, recent body weight fluctuations, arm and calf circumference),the global evaluation (recent acute events, pressure sores, numberof medications taken), dietary indexes (number of mealsconsumed, autonomy in feeding, protein and fiber intake), theobjective evaluation (health and nutritional status).

(over and under-nutrition): Validation of the JaNuS screening tool,

L.M. Donini et al. / Clinical Nutrition xxx (2013) 1e8 3

On the basis of a comprehensive evaluation nutritional status,patients were classified into one of the following types ofmalnutrition:

a) Undernutrition� Energymalnutrition: leanmusclemass and fat mass depletion[Arm circumference (AC) <20.2 cm in men and <18.6 cm inwomen; triceps skinfold thickness (TSF) <5.2 mm in men and<9.7 mm in women] with normal values of plasma proteinsand lymphocyte count.24

� Hypoalbuminemic malnutrition: depletion of visceral pro-teins and immunocompetence (albumin levels <31 g/l,transferrin <2 g/l, lymphocyte count <1500 #/mm3) in thepresence of normal anthropometric parameters.25

� Calorieeprotein malnutrition: when the two above-mentioned conditions coexisted.

b) Overnutrition� Obesity without complications: patients having a fat mass>25% in men and >35% in women, without complications.13

� Obesity with functional complications (impaired glucosetolerance, type 2 diabetes, dyslipidemia).

� Obesity with organic complications (history of coronary heartdisease, stroke) and/or respiratory complications (COPD,OSAS), with or without metabolic complications.

2.1. Second phase

independent explanatory variables (simple parameters to beused by non-specialist personnel to identify the presence ofnutritional risk) were dichotomized and coded as 0 (absent) or 1(present), and divided into two groups: variables related to un-dernutrition or to overnutrition, respectively, derived from MNA(Table 1). Two variables (comorbidity and C-reactive protein) wereconsidered as linked to both undernutrition and overnutrition.

2.1.1. Model constructionAfter verification of normal data distribution, a univariate

analysis was performed using t-test and c2 test in order to inves-tigate the relationship between independent variables and the

Table 1Variables associated with risk of malnutrition.

Overnutrition Undernu

- Family history of T2DM, CV disease, dyslipidemia- BMI >30 kg/m2;- WC >88 cm (F) or >102 cm (M);- Dyslipidemia (total CHOL >200 mg/dl; triglycerides >150 mg/dl;HDL-CHOL <45 mg/dl (M) or <50 mg/dl (F); LDL-CHOL >130 mg/dl);

- Prediabetes (IFG or IGT) or T2DM;- Arterial hypertension;- Respiratory diseases (COPD, OSAS)- Eating behavior disorder (SCOFF � 2)

- BMI <1- Calf cir- Arm ci- Album- Hemog- Total C- Class C- Age >7- Pressur- Medium- Reduce- Need h- Difficu- <2 me- Milk an- Meat o- Eggs or- Fruit an- Reduce- Weigh- Weigh

Legend: ADL, activities of daily living; BMI, body mass index; CHOL, cholesterol; COPD, cglucose tolerance; MNA, Mini Nutritional Assessment; OSAS, obstructive sleep apneaQuestionnaire; WC, waist circumference.

Please cite this article in press as: Donini LM, et al., Risk of malnutritionClinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.12.001

presence of undernutrition or overnutrition (outcome variables).The predictors of outcome variables at univariate analysis wereincluded in a logistic regression model. In order to avoid the con-founding effect of collinearity (verified through the Pearson’s test,t-test or c2 test), a few variables with a similar biologic meaningwere excluded, when correlated with each other.

Logistic regression models were elaborated using a block pro-cedure: all the variables correlated to the outcomes at univariateanalysis were introduced in the model. To evaluate the validity ofthe logistic model the following parameters and procedures weretested:

� the predictive value of the test, by assessing the efficacy (abilityto classify correctly healthy subjects and malnourished pa-tients), sensitivity (ability to identify malnutrition), specificity(ability to identify healthy subjects), positive or negative predic-tive values (reliability of a positive or negative test result,respectively) for different threshold values;

� the verisimilitude test (�2LL), reflecting the difference betweenobserved and expected data;

� the area under the Receiver Operating Characteristics Curve(AUROC), i.e., the graphic representation (varying from 0.5 to1.0) of the sensitivity (probability of Type I errore false negativerate) with respect to the complement of specificity [1 e speci-ficity] (probability of Type II error e false positive rate) of thelogistic model by varying the decision threshold. The AUROCrepresents the probability of a subject to be correctly classifiedby the logistic model.

On the basis of all tested variables, a screening test of malnu-trition was finally constructed, to account for both types ofmalnutrition.

2.2. Third phase: validation of the model

The JaNuS test was administered to each subject belonging tothe validation sample, who also underwent a thorough nutritionalevaluation (clinical anamnesis, anthropometry and biochemicalparameters). Later, the predictive validity of the test was assessed

trition Combined malnutrition

8.5 kg/m2;cumference <31 cm;rcumference <22 cm;in levels <3.5 g/dl;lobin <12 g/dl (F),<13 g/dl (M);HOL <150 mg/dl;frailty;5 years;e sores grade 3 or 4;-severe cognitive impairment;

d autonomy level (>2 functions lost in ADL);elp at meals (MNA)lties at meals (MNA);als/day (MNA);d dairy products <1 portion/day (MNA);r fish products <1 portion/day (MNA);legumes <2/week (MNA);d vegetables <2 portions/day (MNA);d appetite (MNA);t loss >5% in 3 months (MNA);t loss >10% in 6 months (MNA);

- Comorbidity grade 3 or 4;- C-reactive protein >20 mg/l

hronic obstructive pulmonary disease; IFG, impaired fasting glucose; IGT, impairedsyndrome; T2DM, type 2 diabetes mellitus; SPMSQ, Short Portable Mental Status

(over and under-nutrition): Validation of the JaNuS screening tool,

Table 3Nutritional status in the evaluation sample (%), with special reference to undernu-trition in the presence of obesity.

Total (%) Males (%) Females (%)

Nutritional statusNormal 21.1 16.1 23.2Undernutrition 12.2 13.2 12.1Overnutrition 51.6 54.6 50.5Mixed malnutrition 15.1 16.1 14.2

Undernutrition (including mixed malnutrition)Calorie 2.1 5.1 1.3Hypoalbumenic 17.1 14.3 18.7Calorie-protein 8.1 10.9 6.3

Overnutrition (including mixed malnutrition)Uncomplicated 30.8 19.1 37.4Complicated 28.4 42.4 20.6Complicated by COPD, CVD 7.5 9.2 6.7

Undernutrition in the presence of obesityTotal cases 15.9 14.1 14.9Uncomplicated obesity 1.1 4.4 3.5Complicated obesity (T2DM,dislipidemia) in the presenceof undernutrition

9.1 5.3 6.5

Complicated obesity (COPD, CVD)in the presence of undernutrition

5.7 4.4 4.9

Legend: COPD, chronic obstructive pulmonary disease; CV, cardiovascular diseaseT2DM, type 2 diabetes mellitus.

Table 4Variables associated with undernutrition in the evaluation sample.

Undernutritionpresent(n ¼ 66)

Undernutritionabsent(n ¼ 330)

p

Age (years) 71.9 � 15.8 58.0 � 16.5 <0.0001Body mass index (kg/m2) 36.9 � 6.2 41.6 � 9 <0.0001Arm circumference (cm) 27.9 � 7 36.4 � 6.2 <0.0001Calf circumference (cm) 31.4 � 4.9 37.9 � 6.1 <0.0001

L.M. Donini et al. / Clinical Nutrition xxx (2013) 1e84

by calculating the efficiency, sensitivity, specificity, positive andnegative predictive values. For assessing testeretest reliability,Spearman’s rank order correlation (rs) between the first and thesecond administration of the JaNuS was used. A value of rs �0.80was set as the minimum level of acceptable reliability. Wilcoxon’ssigned-rank test was used to test the difference between test andretest scores. Internal consistency of the test was assessed usingCronbach’s a. a values >0.6 may be acceptable, while values >0.8indicate good reliability.

Differences were considered to be statistically significant for p<0.05. Statistical analysis was performed using SPSS 10.0 statisticalsoftware (SPSS Inc, Wacker Drive, Chicago, IL, USA) and WinEpiscope 2.0 [Facultad de Veterinaria di Saragozza (Espana),Wageningen University (the Netherlands), University of Edimburgh(United Kingdom)].

The study protocol was approved by the Ethical Committee ofthe “Sapienza” University of Rome and all subjects gave their oraland written informed consent to the anonymous use of personaldata.

3. Results

3.1. Phase 1: selection of variables

The clinical and functional characteristics of the 396 studyparticipants (131 men and 265 women, mean age 60.6 � SD 17.0years and 62.6 � 18.0 years, respectively) are summarized inTable 2. The recruitment procedure consisted of sequential enroll-ment by admission. The only exclusion criteria were lack ofconsensus and a severely impaired cognitive status.

Comorbidity grade was 1 or 2 in 94.4% of cases, with type 2diabetes, dyslipidemia and hypertension as the most prevalentcomorbid conditions.

The distribution of patients according to the nutritional statusis reported in Table 3. 52% of subjects had signs of overnutrition,accompanied by complications in more than half of cases; 12%had signs of undernutrition, whereas the copresence of over andundernutrition was found in 15% of subjects, with characteristicsof both overnutrition and undernutrition (Tables 3e5), withundernutrition being also observed in several subjects withobesity.

Table 2Characteristics of subjects included in the evaluation sample (% of cases).

Variable Total(n ¼ 396)

Males(n ¼ 131)

Females(n ¼ 265)

Comorbidity class (1/2/3/4) 44.2/50.2/5.4/0.2

43.3/48.2/8.5/0

44.7/50.2/5.4/0.2

Altered glucose regulationImpaired fasting glucose 3.4 5.1 2.5Impaired glucose tolerance 3.1 4.3 1.9Type 2 diabetes 29.3 29.7 29.1Fasting insulin �20 U/l 10.6 11.1 9.7

Altered lipid metabolismTotal cholesterol >200 mg/dl 39.8 29.3 44.9Triglycerides >150 mg/dl 33.4 36.8 35.3LDL-cholesterol >130 mg/dl 30.4 21.4 37.0HDL-cholesterol <45 mg/dl(M), <50 (F)

68.9 70.2 69.4

Arterial hypertension (%) 65.0 66.9 64.2Respiratory disease (COPD, OSAS) 31.5 42.2 26.3Pressure sores grade 3e4 9.2 10.3 8.7Lost activities of daily living >2 20.4 19.4 20.0Medium-severe cognitive decline

(SPMSQ)11.5 12.7 11.0

Legend: COPD: chronic obstructive pulmonary disease, OSAS: obstructive sleepapnea syndrome, SPMSQ: Short Portable Mental Status Questionnaire.

Please cite this article in press as: Donini LM, et al., Risk of malnutritionClinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.12.001

3.2. Phase 2: test construction

Two groups of variables were selected for their significant cor-relation with the outcome variables:

- 12 variables were related to undernutrition: calf circumference<31 cm; arm circumference <22 cm; albuminemia <3.5 g/dL;

Albumin levels (g/dl) 3.2 � 0.4 3.8 � 1.8 <0.0001Hemoglobin (g/dl) 11.8 � 1.7 13.4 � 1.5 <0.0001C-reactive protein (mg/l) 25.4 � 35.4 8.2 � 8.6 <0.0001Total cholesterol (mg/dl) 160.1 � 36.6 200 � 36.6 <0.0001Comorbidity grade 3e4 (%) 17.8 1.5 <0.0001Lost activities of daily living >2 (%) 49.6 10.8 <0.0001Body mass index <18.5 kg/m2 (%) 11.8 0.4 <0.0001Arm circumference �22 cm (%) 14.2 0 <0.0001Calf circumference <31 cm (%) 42.7 10.3 <0.0001Albumin levels �3.5 g/dl (%) 80.1 18.7 <0.0001Hemoglobin <12 g/dl (F),

<13 (M) (%)63.1 21.1 <0.0001

C reactive protein >20 mg/l 40.6 9.0 <0.0001Cholesterol levels �150 mg/dl 41.6 8.5 <0.0001Medium-severe cognitive

impairment (%)29.0 6.0 <0.0001

Weight loss >5% in 3 months 27.0 11.9 <0.0001Weight loss >10% in 6 months 26.6 8.7 <0.0001Pressure sores grade 3e4 (%) 24.1 4.4 <0.0001Need help at meals (%) 24.8 3.5 <0.0001Difficulties at meals (%) 30.7 6.1 <0.0001Reduced appetite (%) 47.2 7.8 <0.0001<2 meals/day (%) 14.8 1.9 <0.0001Milk and dairy products <1

portion/day22.0 14.6 <0.0001

Meat or fish <1 portion/day 40.4 12.7 <0.0001Eggs or legumes <2 portions/week 31.2 10.7 <0.0001Fruit or vegetables <2 portions/day 27.5 7.8 <0.0001

(over and under-nutrition): Validation of the JaNuS screening tool,

Table 5Variables associated with overnutrition.

Overnutritionpresent(n ¼ 279)

Overnutritionabsent(n ¼ 117)

p

Age (years) 60.7 � 16.83 74.9 � 14.1 <0.0001Body mass index (kg/m2) 41.6 � 9.5 19.7 � 3.8 <0.0001Waist circumference (cm) 120.2 � 16.9 77.7 � 11.8 <0.0001C-reactive protein (mg/l) 12.2 � 18.4 23.6 � 29.0 0.001Fasting glucose (mg/dl) 107.1 � 42.4 91.1 � 34.6 0.018Fasting insulin (mU/ml) 16.4 � 15.4 4.4 � 2.7 <0.0001Total cholesterol (mg/dl) 193.5 � 43.5 157.6 � 33.8 <0.0001Triglycerides (mg/dl) 153.3 � 81.2 107.5 � 62.2 0.001Comorbidity grade 3e4 (%) 3.6 25.0 0.000Family history of diseaseCardiovascular disease (%) 47.9 44.4 0.445Type 2 diabetes mellitus (%) 55.3 33.3 0.022Dyslipidemia (%) 23.6 3.7 0.008One or more (%) 78.9 59.3 0.022

Arterial hypertension (%) 69.3 42.9 0.001Respiratory diseases (COPD, OSAS) 32.2 33.3 0.509Body mass index �30 kg/m2 (%) 90.3 0 0.000Waist circumference �88 cm (F),

�102 (M) (%)99.4 22.2 0.000

Altered glucose regulationImpaired fasting glucose (%) 3.6 3.4 0.07Impaired glucose tolerance (%) 2.8 0Type 2 diabetes mellitus (%) 35.1 27.6Fasting insulin>20 U/l (%) 13.8 0

Altered lipid metabolism 85.4 81 0.286PCR � 20 mg/l 16 38.5 0.001Eating behavior disorders

(SCOFF � 2)11.1 0 0.018

L.M. Donini et al. / Clinical Nutrition xxx (2013) 1e8 5

hemoglobin <12 g/dL in women and <13 in men; total choles-terol level < 150 mg/dL; PCR >20 mg/L; age �75; presence of ahigh comorbidity level (3 or 4); pressure sores grade III or IV;moderate or severe cognitive impairment; need of assistance atmeals; consumption of less than two meals per day;

- 7 variables correlated to overnutrition: positive family historyfor type 2 diabetes; BMI�30 kg/m2; waist circumference�88 inwomen or 102 cm in men; impaired lipid profile (total choles-terol level >200 mg/dL; triglyceride level >150 mg/dL; HDL-cholesterol < 45 mg/dL in men or < 50 in women; LDL-cholesterol level >130 mg/dL); type 2 diabetes; hypertension;eating disorders (SCOFF � 2);

A few variables with a similar biologic meaning, highly corre-lated with each other, were excluded to avoid the confoundingeffect of collinearity (tested with Pearson’s r, t-test or c2). The finalJaNuS test consists of 19 items.

These variables were evaluated in different logistic models forover- and under-nutrition, respectively (Table 6). The logistic modelfor overnutrition showed a high efficacy (97.7%), with high sensi-tivity and specificity. The value of the verisimilitude test of the finalmodel was lower than the value of the model with only one con-stant. Cox & Snell’s R2 and Nagelkerke’s R2 showed that the varianceexplained by the logistic model increased when the explanatory

Table 6Predictive value of logistic models for malnutrition (overnutrition and undernutrition) in

Model Predictive value �2LL R

EFF SENS SPEC PPV NPV C

Overnutrition 97.7 99.6 66.6 99.2 80.0 2.8 (initial, 97.5) 0Undernutrition 84.4 83.6 84.8 75.3 90.3 131.3 (initial, 205.9) 0

Legend: AUROC, area under the Receiver Operating Characteristics Curve (Null hypothesnegative predictive value; PPV, positive predictive value; SE, standard error; SENS, sensi

Please cite this article in press as: Donini LM, et al., Risk of malnutritionClinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.12.001

variables were added to the constant variable. The AUROC (equal to0.999) confirmed the validity of the logistic model in correctlyclassifying the cases.

The logistic model for undernutrition showed a high efficacy(84.4%) and high sensitivity and specificity. The value of the �2LLtest of the final model was lower than the value in the model withonly one constant. Cox & Snell’s R2 and Nagelkerke’s R2 showed thatthe variance explained by the logistic model was significantlyincreased. Also in this case, the area under the ROC curve (equal to0.926) confirmed the validity of the logistic model in classifyingcorrectly the subjects.

These logistic models were used to construct a test composed oftwo parts: the overnutrition and the undernutrition subtest, con-sisting of 7 and 12 items, respectively (Fig. 1). The test was namedJaNuS (Just A Nutritional Screening), also referring to JaNuS, thetwo-faced Roman God, in order to highlight its main characteristic,i.e., the possibility to explore both aspects of malnutrition.

Each item was scored on the basis of its major or minor corre-lation with the outcome variable. In particular the score wasattributed considering, at the logistic model, the estimated oddsratio of the variable which tells changes in odds for a case when thevalue of that variable increases by 1. The scores were weighed inorder to obtain a maximum score of 20 in both parts of the test. Todefine the positivity of the test, a threshold value of 5 was estab-lished for both the JaNuS defect and the JaNuS excess subtests: thisvalue corresponded to the best value of sensitivity, specificity, aswell as the best positive predictive value. Hence, a score >5 indi-cated the presence of overnutrition or undernutrition dependingon the section of the test.

Taking into account the frequent coexistance of both the types ofmalnutrition, a graphic model was developed using a two-dimension Cartesian coordinate system (Fig. 2). The x-axis is thescore of the JaNuS overnutrition section, whereas the y-axis is thescore of the JaNuS undernutrition section. The length of the resul-tant vector reflects the global level of malnutrition. The angle be-tween the x-axis and the resultant vector (a) represents the level ofco-expression of the two types of malnutrition: an angle equal to0� indicates that only overnutrition is present; an angle equal to 90�

corresponds to isolated undernutrition. This angle can be calcu-lated in radiants using the inverse trigonometric function arctana ¼ undernutrition/overnutrition. To obtain the degrees of theangle, this value must be multiplied by 180 and divided by p.

3.3. Phase 3: validation test

In the validation sample 73 subjects were included (24 men, 49women; mean age, 58.2 � 17.0 years and 60.2 � 18.0, respectively).The distribution of weight and malnutrition was similar to that ofthe estimation sample, with 20.5% of cases being normal weightand with a normal nutritional status, 24.7% with undernutrition,41.1 with overnutrition (excluding overweight) and 13.7% withcombined malnutrition.

The predictive value of the JaNuS test for overnutrition andundernutrition in the validation sample is reported in Fig. 3. Both

the estimation sample.

2 AUROC

ox & Snell Nagelkerke

.42 (initial, 0.21) 0.95 (initial, 0.77) 0.999 (SE, 0.001; 95% CI, 0.997e1.002)

.49 (initial, 0.27) 0.67 (initial, 0.37) 0.926 (SE, 0.019; 95% CI, 0.89e0.96)

is, area ¼ 0.5); CI, confidence interval; EFF, efficacy; �2LL, �2 Log likelihood; NPV,tivity; SPEC, specificity.

(over and under-nutrition): Validation of the JaNuS screening tool,

Fig. 1. JaNuS (Just A Nutritional Screening) test.

L.M. Donini et al. / Clinical Nutrition xxx (2013) 1e86

scales had a high predictive value (>80%). Sensitivity was higher forthe JaNuS test for overnutrition (95%), whereas the JaNuS test forundernutrition had a very good specificity (86.7%). Cronbach’s co-efficient was higher than the suggested minimum value of 0.70while testeretest variability was acceptable (rs> 0.80). Statisticallysignificant differences were never observed between the first andthe second test.

4. Discussion

The study validates a screening tool for the assessment of thenutritional risk in terms of both over- and undernutrition. This toolis characterized by: a) parameters easy-to-use by the health carepersonnel lacking of nutritional competence; b) excellent predic-tive validity. For these reasons, it might be confidently used in the

Please cite this article in press as: Donini LM, et al., Risk of malnutritionClinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.12.001

clinical setting to determine the importance of malnutrition (anytype, including sarcopenic obesity) in the risk of morbidity andmortality.

Although a variety of screening tests has been developed forundernutrition, mainly in the elderly,5 they are largely unsatisfac-tory. In Geriatrics, as well as in Surgery and Critical care settings,undernutrition is associated with progressive worsening of theclinical status (reduced immune response, poor surgical woundrepair and skin pressure sore healing), with reduced autonomy andimpairment of the cognitive status.2,3 Undernutrition may be pre-sent also in subjects with BMI higher than the lower limit of normalweight (18.5 kg/m2) and the biochemical correlates of undernu-trition may be biased in a variety of clinical conditions.26

Also overnutrition may be incorrectly identified. Excess body fatand visceral adiposity have been widely recognized as conditions

(over and under-nutrition): Validation of the JaNuS screening tool,

Fig. 2. Graphic model of the JaNus test.

L.M. Donini et al. / Clinical Nutrition xxx (2013) 1e8 7

with a negative impact on cardio-metabolic morbidity and mor-tality. Early diagnosis and intervention are mandatory, as the longerthe duration of obesity, the higher the health risk.27 Despiteepidemiological evidence, the very high prevalence of obesityconstitutes a barrier to diagnosis and treatment. Obesitymay be leftunrecognized as a relevant source of clinical problems among pa-tients admitted in to cardiac or respiratory rehabilitation units.Clinicians’ attention is often directed only to the acute clinicalevents leading to admission, and poor consideration is given to theunderlying cause. Excess body fat is also linked to an increased riskof disability,28 leading to impaired quality of life. In most cases, thepresence of BMI �30 kg/m2 is associated with increased fat mass,but the problem of copresence of over and undernutrition deservesmore attention. BMI per se explains only the 2.1% of the variance offat mass.

Fig. 3. Predictive value of the JaNuS test for ovenutrition and undernutrition in thevalidation sample.

Please cite this article in press as: Donini LM, et al., Risk of malnutritionClinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.12.001

Of note, it is now well accepted that the two conditions of over-and undernutrition may frequently coexist, particularly in theelderly and in obese subjects, where sarcopenic obesity may ensue,due to the aging process, inappropriate dieting, and the conse-quences of bariatric surgery.29 In this clinical condition, the twoaspects of malnutrition not only coexist, but act synergistically in asort of vicious circle. Sarcopenia reduces energy expenditure, fa-voring fat mass accumulation; on the other hand, fat mass pro-motes the synthesis of inflammatory factors and cytokines, favoringthe loss of muscle fibers.27 These mechanisms may lead toincreased morbidity and disability, to poor quality of life and toincreased mortality.

Sarcopenic obesity may be common in the population, alsodepending on the prevalence of obesity. In the Korean sarcopenicobesity study (KSOS), involving 2221 Koreans over 60 yearsenrolled into the Fourth Korea National Health and Nutrition Ex-amination Survey, the prevalence rates of sarcopenic obesity was6.1% in males and 7.3% in females, and was positively associatedwith the number of combined medical conditions, as well as withlow physical activity.30 An additional study identified sarcopenicobesity in 18.4% of males and 25.8% of females and the sarcopenicobese group was significantly associated with insulin resistance,metabolic syndrome, and cardiovascular disease risk factors.31 Inadults aged �70, subjects with sarcopenia and obesity, eitherindependently or concurrently, also had poorer cognitive func-tioning compared to non-sarcopenic non-obese older adults.32

These data have important clinical consequences: any effortshould be done to treat sarcopenic obesity, thus reducing the risk ofadverse events.33

Thus, an ideal test should be able to make a comprehensiveassessment of under- and over-nutrition, based on variables easilymeasurable by health care personnel, also lacking nutritionalcompetence (nurses, caregivers, physicians usually working inother clinical fields); the test should have high predictive value,high sensitivity and reliability.

In the JaNuS test all these requirements were fulfilled. The finalJaNuS test consists of 19 items, a number larger than other nutri-tional screening tools. This large number is justified by the need ofcomprehensive assessment, including both under- and over-nutrition in a single test. Moreover, most of the selected variablesmay be already present in the clinical or nurse folders (laboratorytests, pressures sores, therapy, blood pressure) and the bed-sidemeasurements are easy and rapid to perform (anthropometricmeasurements) by caregivers non-specialized in nutrition. Whennot available, the biochemical parameters can be analyzed atreasonably low cost. The whole test requires about 15 min, anddemonstrated a good predictive value, a high sensitivity and a goodspecificity.

Several limitations of the JaNuS test should however beacknowledged. First, it was developed and validated in a pre-geriatric population; this explain why the cutoff values of a fewparameters were derived from the geriatric population of the MNAtest.9 Hence its validity should be verified for younger age groups,or different cut-offs should be considered. Selective use of param-eters in specific populations may be hypothesized. Second, a po-tential limitation of the tool concerns some of the items used toestablish the reference criteria for nutritional status: low albuminconcentration and low lymphocyte count in particular. These pa-rameters, although they may be affected by a range of conditionsunrelated to nutritional status, are however habitually includedamong parameters utilized for nutritional assessment, in particularin clinical practice.34 Finally, we applied equations using anthro-pometric data for the assessment of body composition. These pa-rameters show a relatively high inter- and intra-operatorvariability, and are considered to have a good usefulness in clinical

(over and under-nutrition): Validation of the JaNuS screening tool,

L.M. Donini et al. / Clinical Nutrition xxx (2013) 1e88

practice, but technology may offer new parameters more closelyassociated with outcome variables.

In conclusion, this validation study of the JaNuS test indicatesthat it may represent an easy-to-use tool for the early identificationof an altered nutritional status in terms of undernutrition, over-nutrition or copresence of both the conditions. Further studies areneeded to verify and confirm its external validity for a widespreaduse in the clinical setting.

Conflict of interest

The authors declare no conflict of interest.

Acknowledgments

LMD designed the research; LMR and BN conducted theresearch; LMD, LMR, EP and GM wrote the manuscript.

References

1. Kondrup J, Allison SP, Elia M, Vellas B, Plauth M. ESPEN guidelines for nutritionscreening 2002. Clin Nutr 2003;22:415e21.

2. Cruz-Jentoft AJ, Baeyens JP, Bauer JM, Boirie Y, Cederholm T, Landi F, et al.Sarcopenia: European consensus on definition and diagnosis: report of theEuropean Working Group on sarcopenia in older people. Age Ageing 2010;39:412e23.

3. Bales CW, Ritchie CS. Sarcopenia, weight loss, and nutritional frailty in theelderly. Annu Rev Nutr 2002;22:309e23.

4. Elia M. Obesity in the elderly. Obes Res 2001;9(Suppl. 4):244Se8S.5. Davidson I, Smith S. Nutritional screening: pitfalls of nutritional screening in

the injured obese patients. Proc Nutr Soc 2004;63:421e5.6. Green SM, Watson R. Nutritional screening and assessment tools for older

adults: literature review. J Adv Nurs 2006;54:477e90.7. Heersink JT, Brown CJ, Dimaria-Ghalili RA, Locher JL. Undernutrition in hospi-

talized older adults: patterns and correlates, outcomes, and opportunities forintervention with a focus on processes of care. J Nutr Elder 2010;29:4e41.

8. Shikora SA, Jenson GL. Hypoenergetic nutrition support in hospitalized obesepatients. Am J Clin Nutr 1997;66:679e80.

9. Guigoz Y, Vellas B, Garry PJ. Assessing the nutritional status of the elderly: theMini Nutritional Assessment as part of the geriatric evaluation. Nutr Rev1996;54:S59e65.

10. Malnutrition Advisory Group. A consistent and reliable tool for malnutritionscreening. Nurs Times 2003;99:26e7.

11. Kondrup J, Rasmussen HH, Hamberg O, Stanga Z. Nutritional risk screening(NRS 2002): a newmethod based on an analysis of controlled clinical trials. ClinNutr 2003;22:321e36.

12. Covinsky KE, Martin GE, Beyth RJ, Justice AC, Sehgal AR, Landefeld CS. The rela-tionship between clinical assessments of nutritional status and adverse out-comes in older hospitalized medical patients. J Am Geriatr Soc 1999;47:532e8.

13. De Lorenzo A, Deurenberg P, Pietrantuono M, Di Daniele N, Cervelli V,Andreoli A. How fat is obese? Acta Diabetol 2003;40(Suppl. 1):S254e7.

Please cite this article in press as: Donini LM, et al., Risk of malnutritionClinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.12.001

14. American Diabetes Association. American Diabetes Association: clinical prac-tice recommendations 2002. Diabetes Care 2002;25(Suppl. 1):S1e147.

15. Expert Panel on Detection, Evaluation, and Treatment of High Blood Choles-terol in Adults. Executive summary of the third report of the NationalCholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation,and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III).J Am Med Assoc 2001 May 16;285(19):2486e97.

16. Greenfield S, Apolone G, McNeil BJ, Cleary PD. The importance of co-existentdisease in the occurrence of postoperative complications and one-year re-covery in patients undergoing total hip replacement. Comorbidity and out-comes after hip replacement. Med Care 1993;31:141e54.

17. Shea JD. Pressure sores: classification and management. Clin Orthop Relat Res1975:89e100.

18. Pfeiffer E. A short portable mental status questionnaire for the assessment oforganic brain deficit in elderly patients. J Am Geriatr Soc 1975;23:433e41.

19. Katz S, Downs TD, Cash HR, Grotz RC. Progress in development of the index ofADL. Gerontologist 1970;10:20e30.

20. Morgan JF, Reid F, Lacey JH. The SCOFF questionnaire: assessment of a newscreening tool for eating disorders. BMJ 1999;319:1467e8.

21. Lohman TJ, Roache AF, Martorell R. Antropometric standardization referencemanual. Med Sci Sports Exerc 2002;24:952.

22. Durnin JV, Womersley J. Body fat assessed from total body density and itsestimation from skinfold thickness: measurements on 481 men and womenaged from 16 to 72 years. Br J Nutr 1974;32:77e97.

23. Siri WE. Body composition, from fluid spaces and density. Analysis of methods.In: National Academy of Sciences, editor. Techniques for measuring bodycomposition. Washington, D.C.: National Research Council; 1963.

24. de Groot LC, Hautvast JG, van Staveren WA. Nutrition and health of elderlypeople in Europe: the EURONUT-SENECA Study. Nutr Rev 1992 Jul;50(7):185e94.

25. Linee Guida SINPE per la Nutrizione artificiale ospedaliera. Riv It Nut Parent Ent2002:S5.

26. Don BR, Kaysen G. Serum albumin: relationship to inflammation and nutrition.Semin Dial 2004;17(6):432e7.

27. Zamboni M, Mazzali G, Fantin F, Rossi A, Di Francesco V. Sarcopenic obesity: anew category of obesity in the elderly. Nutr Metab Cardiovasc Dis 2008;18:388e95.

28. Visser M. Obesity, sarcopenia and their functional consequences in old age.Proc Nutr Soc 2011;70:114e8.

29. Roubenoff R. Sarcopenic obesity: the confluence of two epidemics. Obes Res2004;12:887e8.

30. Hwang B, Lim JY, Lee J, Choi NK, Ahn YO, Park BJ. Prevalence rate and associatedfactors of sarcopenic obesity in Korean elderly population. J Korean Med Sci2012;27:748e55.

31. Chung JY, Kang HT, Lee DC, Lee HR, Lee YJ. Body composition and its associationwith cardiometabolic risk factors in the elderly: a focus on sarcopenic obesity.Arch Gerontol Geriatr 2013;56:270e8.

32. Levine ME, Crimmins EM. Sarcopenic obesity and cognitive functioning: themediating roles of insulin resistance and inflammation? Curr Gerontol GeriatrRes 2012;2012:826398.

33. Mathus-Vliegen EM. Prevalence, pathophysiology, health consequences andtreatment options of obesity in the elderly: a guideline. Obes Facts 2012;5:460e83.

34. Numeroso F, Barilli AL, Delsignore R. Prevalence and significance of hypo-albuminemia in an internal medicine department. Eur J Intern Med 2008;19(8):587e91.

(over and under-nutrition): Validation of the JaNuS screening tool,