PhilSPEN Online Journal of Parenteral and Enteral Nutrition

(Article 23 | POJ_0016.html) Issue February 2012 - December 2014: 57-73

Original Clinical Investigation

Pediatric Nutrition Assessment Validation Study: Report from the Philippines

Abstract | Introduction | Methodology | Results | Discussion | Conclusion | References | PDF (756 KB) |Back to Articles Page

Submitted: March 12, 2014 | Posted: August 10, 2014


Grace Paguia* MD, Luisito Llido MD

*Corresponding Author; Email:


Pediatric Clinical Nutrition Program, Clinical Nutrition Service, St. Luke’s Medical Center, Quezon City, Metro-Manila, Philippines



Background: There is no pediatric nutrition assessment tool used in the Philippines thus the Pediatric Clinical Nutrition Program of the Clinical Nutrition Service decided to create one and validate it in order to fulfill the requirement of the clinical nutrition process for pediatric patients in this institution.

Objective: To determine the sensitivity, specificity, predictive values, likelihood ratio (LR), ROC (Receiver Operating Characteristic) curves and AUC (Area Under the Curve) of the SLMC nutrition assessment tool/form for pediatrics in identifying malnourished children including those at risk for malnutrition and its related complications.

Methodology: A total of 214 pediatric social service patients (in patient and outpatient) that were referred to the clinical nutrition service physicians at St. Luke’s Medical Center, Quezon City (January 2012 to January 2014) were assessed along with well patients seen at the Pedia Day activity (October 2013). The pediatric nutrition assessment form/tool was used.  The validity of the assessment tool was then analyzed using the following statistical tools: sensitivity, specificity, positive and negative predictive values, ROC curves and AUC. The following components of the tool were analyzed: BMI, SGA, nutritional status based on combined criteria, serum albumin, Total Lymphocyte Count (TLC), Nutrition Risk Score (NRS), Weight and Height for Age and Length for Age.

Results: The following components were highly specific for severe malnutrition: SGA-C (93.2% LR=2.6), Combined criteria - severe malnutrition (86.3%, LR=0.54), Albumin <2.1g (86.3%, LR=0.54), TLC <1000 (98.3, LR=4.3), and Nutrition Risk Score – High Risk (97.5%, LR=4.9). The following are highly sensitive for normal to mild malnutrition: BMI – normal (89.5%), SGA-A (99%), SGA-B (81.1%), Combined criteria – normal (99.7%), Albumin > 3g (99.7%), NRS – mild risk (99.7%), Length for Age – normal (83.3%). Positive Predictive Value (PPV) for malnutrition was high for SGA-C (64.4%), TLC <1000 (75%), and NRS – high risk (77.2%).

Conclusion: The pediatric nutrition assessment form/tool designed by the Clinical Nutrition Service, Pediatric Clinical Nutrition Section, is an acceptable tool for detecting the presence or absence of malnutrition in pediatric patients.


KEYWORDS: Pediatrics, nutrition assessment, SGA, BMI, anthropometric, malnutrition



According to the World Health Organization, malnutrition is essentially “bad nourishment”.  It includes too little as well as excessive food intake, wrong type of food and the disease states/complications that come along with it. (1) Clinically, it is characterized by inadequate or excess intake of carbohydrate, fat, and protein with the accompanying complications of frequent infections and metabolic disorders. People are malnourished if they are unable to utilize fully the food they eat i.e. due to diarrhea or other illness (secondary malnutrition), if they consume too many calories (overnutrition), or if their diet does not provide adequate calories and protein for growth and maintenance (undernutrition or protein-energy malnutrition: WHO, World Water Day 2001). (2)

Malnutrition in all its forms increases the risk of disease and early death. Protein-energy malnutrition, for example, plays a major role in half of all under-five deaths each year in developing countries (WHO 2000). (3) In pediatric patients, the WHO has recommended guidelines for the use of growth charts to monitor development and identify those who are falling outside the normal curve. (4) Such cases are identified using z-score growth charts (WHO growth charts 2006) specific for age, gender, weight, height (for age) and weight for height.  However these charts do not contain information as to the cause of the growth delay nor do they indicate the risk for complications.  Furthermore, there is no uniform  nutrition assessment tool/assessment form for the pediatric age group (A.S.P.E.N. working group on Defining Pediatric Malnutrition, 2013). (5)

In the Philippines the pediatric societies used nutrition screening tools designed by local agencies like the FNRI (Food and Nutrition Research Institute) and the WHO (World Health Organization). (6,7) There was, however, no nutrition assessment tool except for one developed for the Clinical Nutrition Fellowship Training Program in St. Luke’s Medical Center and cited by the Philippine Society of Parenteral and Enteral Nutrition (PhilSPEN). When the JCIA (Joint Commission International) standards accreditation was invited to assess the clinical nutrition process in this institution (St. Luke’s Medical Center) the need to validate the tool became a priority goal. (8) In this regard the Clinical Nutrition Service and the Clinical Nutrition Fellowship Training Program, through its first pediatric clinical nutrition fellow, performed the validation process of this pediatric nutrition assessment form/tool (Figure 1) and to make the necessary modification(s) based on the results of the validation process.


Figure 1: Pediatric Nutritional Assessment Form - Initial



The components of the initial pediatric nutrition assessment form (Figure 1) are the following:

  1. Body Mass Index (BMI)
  2. Subjective Global Assessment (SGA) method modified for the pediatric age group. The modifications are: oral motor skill assessment for two years old and below, and feeding tolerance
  3. Weight for Height (wasting)
  4. Length/Height for Age (stunting)
  5. Head circumference (for 0 to three (3) years) – this was removed due to the difficulty of performing an accurate head circumference on the selected age group.
  6. Nutritional Status based on combined criteria of BMI, SGA, weight for height, length/height for age, and head circumference
  7. Albumin
  8. Total Lymphocyte Count (TLC)
  9. Nutrition Risk Score (NRS) based on BMI score, SGA score, Albumin and Total Lymphocyte Count (TLC)

All these elements were subjected to the validation process which are the following: a) sensitivity, b) specificity, c) positive and negative predictive value, d) likelihood ratio (LR), e) ROC (Receiver Operating Characteristic) Curves and f) Area Under the Curve (AUC). (9,10)

The participants of this study consisted of all pediatric social service patients (OPD clinic and in patient) that were referred to the clinical nutrition service physicians from January 2012-January 2014, including patients seen at the Pedia Day activity (October 2013). The clinical nutrition fellow and the clinical dietitian then performed Nutritional Assessment on all referred patients using the nutritional assessment form for pediatrics. The clinical nutrition fellow is a general pediatrician while the clinical dietitian is a registered nutritionist/dietitian. Both primary and secondary malnutrition cases were included in this study.  Age and gender appropriate WHO growth charts were used to plot the weight and height/length, weight for height/length and BMI.  The study did not include neonates (less than 1 month old) and those above 18 years old.  

All the assessment forms were then collected and classified as to true positive, true negative, false positive and false negative.  The Reference Standard for the diagnosis of malnutrition or not was decided and chosen by the senior members of the clinical nutrition team. The Reference Value is the Pediatric Nutrition Assessment Validation Code which was also determined by the senior clinical  nutrition consultants for General Pediatrics. The data were encoded into the research database by a registered clinical nutritionist-dietitian. Statistical analysis was done using the NCSS-PASS© software designed by J. Hintze (


The total number of participants assessed was 214. The age and sex distribution of the patients evaluated is shown in Table 1. The male to female ratio is 1.7 to 1.


These are the results of the validation process for the specific components of the pediatric nutrition assessment form/tool:

1. Body Mass Index (BMI)


        • Area Under the Curve (AUC) = 0.517 (0.396 – 0.62)
        • ROC Curve


2. Subjective Global Assessment (SGA)


      • ROC Curve

3. Pediatric Nutritional Status as determined by all the combined criteria


        • ROC


    4. Albumin


        • ROC


    5. Total Lymphocyte Count (TLC)


        • ROC


6. Over-All Nutrition Risk Score (NRS)


        • ROC


7. Weight and Height for Age:


        • ROC


8. Length and Height for Age


        • ROC



Over-all results of the validation process (Table 10):

  1. BMI – sensitivity is high for detecting normal values and specificity is also high for detecting underweight values. However the likelihood ratio and area under the curve is low. The ROC curve showed a pattern due to chance (9).
  2. SGA – sensitivity is high for normal values (=SGA “A”) and mild malnutrition (=SGA “B”), but the likelihood ratio is higher for mild malnutrition. Specificity is high for severe malnutrition (=SGA “C”) together with the likelihood ratio and PPV. AUC for SGA “C” is also highest and the ROC curve shows a superior diagnostic test (9)
  3. The Pediatric Nutritional Status arrived at from the combined criteria showed high sensitivity for normal nutritional status with low PPV and LR. The specificity is high for severe malnutrition, but this has low PPV and LR. The AUC is slightly above 0.5, but the ROC shows the pattern of an inferior diagnostic test.
  4. Albumin – sensitivity is high for normal albumin level (>3 g/dL) but the PPV and LR are low. Specificity is high for low albumin levels (<2.1 g/dL), but the PPV and LR are also low. AUC is slightly high at 0.58, but the ROC is a mixed pattern of a good diagnostic test when the albumin values are low, but an inferior one when the albumin values are on the normal side.
  5. Total Lymphocyte Count (TLC) – the sensitivity is very low, but the specificity is high for values below 1,000 or “low cellular immunity” with high PPV and LR. The AUC is slightly high at 0.575 and the ROC showed similar pattern to albumin.
  6. The Nutrition Risk Score (NRS) – sensitivity is high for Low Risk status, but the PPV and LR are low. The specificity for “High Risk” is high with both PPV and LR equally high. The AUC is high and the ROC curve is indicative of a superior diagnostic test.
  7. Weight and Height for Age – sensitivity and specificity values are low for all categories of normal to malnourished. The AUC is below 0.5 and the ROC showed the pattern that is due to chance.
  8. Length and Height for Age – sensitivity is high for normal values and specificity is high for “high risk for malnutrition”. The PPV and LR values for both were low. The AUC is at 0.5 and the ROC curve is indicative of values that are due to chance.



More than 70% of children with protein-energy malnutrition live in Asia, 26% live in Africa, and 4% in Latin America and the Caribbean (WHO 2000). (1) In the Philippines, data from the National Nutrition Survey done in 2008 showed that in every 100 pre-school children, 26 were underweight, about 28 were stunted and 6 were wasted. (11) Two (2) in every 100 if the same age group were overweight for age and three (3) were overweight for their height.  The WHO growth standards z-score charts (4) were used to classify children that were included in this survey. Further evaluation of the resulting data indicated that the tool(s) used for the survey could not fully qualify/quantify the nutritional status and the risk of developing nutrition related complications in the different pediatric age groups. Existing validated forms are mainly for screening and these differ per hospital or institution, which led the investigators of this study to conduct a validation of the pediatric nutrition assessment form, which was developed by the clinical nutrition service of St. Luke's Medical Center and has been in existence for some time now.

The pediatric nutrition assessment form was developed out of the need to have a more in-depth nutritional assessment tool which the existing nutrition screening tools could not provide. (6,7) The observation that the SGA (subjective global assessment) for adults provided a good nutritional assessment result prompted the pediatric clinical nutrition section of the clinical nutrition service to adopt this tool for children. The addition of anthropometric data like BMI (Body Mass Index) and laboratory data like albumin and total lymphocyte count (TLC) were also noted to further improve the risk-status leveling of adult patients as shown in a recently validated nutrition assessment tool for adults, the "modified SGA" form. (12,13) These observations were eventually translated into the initial pediatric nutrition assessment form. (Figure 1) This is the final phase of the pediatric nutrition assessment form development - the validation process.

Based on the validation outcomes (Table 10) the SGA is shown to be highly specific for patients classified as SGA "C" (97%) or severe malnutrition. The data yielded an AUC (Area Under the Curve) of 0.726, which demonstrated its strong ability to identify malnourished subjects. On the other hand, SGA’s  sensitivity for SGA "C", however high for SGA "A" and "B", is low at 17%. This observation can be compared with a recent study by Young et al. (14) that showed SGA is better at identifying existing malnutrition in elderly patients. The BMI, Weight for Age and Height/Length for Age when used alone did not prove to be good measures for malnutrition determination. All showed high sensitivity for subjects classified as normal to moderate malnutrition but with low PPV and an AUC of less than 0.5. Interestingly, TLC and serum albumin at levels lower than normal yielded high specificity, PPV and LR making them both important components of the assessment tool for determining the presence of malnutrition. The NRS (Nutrition Risk Score) on the other hand showed the same findings as in the comparative study done by Young et al. Its high specificity (97.5%), PPV (77.2%), LR (4.9) and AUC (0.65) demonstrates its good ability to identify subjects at high risk for malnutrition. Length/Height for Age (WHO growth standards 2006), which was found to be low, may be indicative of chronic malnutrition. Data in this study showed a high sensitivity and specificity for the Length/Height for Age, however, PPV (43.6%), LR (1.11) and AUC (0.5)  showed that it is not a reliable tool to diagnose malnutrition when used alone.

The initial pediatric nutritional assessment form (Figure 1) was modified based on the above results. These components were considered valid and were included: a) SGA, b) Albumin, c) TLC, and d) Nutrition Risk Score. The BMI and combined result of Nutritional Status were also included since they contributed to the over-all Nutrition Risk Score. The following components were removed due to their low results and inferior ROC patterns: Weight for Height and Length/Height for Age. As mentioned earlier, the Head Circumference was not included due to practical reasons – inability to get accurate head circumferences in this age group. The final Pediatric Nutrition Assessment Form is shown in Figure 2.


Figure 2: Pediatric Nutritional Assessment Form - Final | (Download sample JPEG file)


The pediatric nutrition assessment tool designed by the Clinical Nutrition Service, Pediatric Clinical Nutrition Section, is an acceptable tool for the determination of the presence or absence of malnutrition in pediatric patients based on the validation process performed.



  1. WHO: Malnutrition. Available at: child/malnutrition/en/. Accessed July 29, 2014.
  2. WHO: World Water Day 2001. Available at: Accessed July 30, 2014.
  3. WHO: Children: reducing mortality. Available at: Accessed July 31, 2014.
  4. WHO Child Growth Standards. Available at: Accessed July 31, 2014.
  5. ASPEN Working Group on Pediatric Malnutrition: Defining Pediatric Malnutrition: A Paradigm Shift Toward Etiology-Related Definitions. JPEN J Parenter Enteral Nutr published online 25 March 2013. Available at: Defining_Pediatric_Malnutrition_American-Society_Parenteral_Enteral_Nutr-2013.pdf. Accessed July 31, 2014.
  6. Llido EP et al. Comparison of standard values of nutrition screening and assessment using BMI percentiles from FNRI-PPS, IRS, CDC 2000, and WHO-MGRS child growth standards in the pediatric population of a tertiary care hospital in the Philippines admitted between years 2000 and 2003. PhilSPEN Online Journal of Parenteral and Enteral Nutrition. Available at: Accessed July 31, 2014.
  7. Llido EP, Aquino C, Santos MA, Llido LO. The comparison between percentile and z-score in the BMI based nutrition screening of pediatric patients in the out-patient department of the Institute of Pediatrics in St. Luke’s Medical Center, Quezon City, Metro-Manila, Philippines. PhilSPEN Online Journal of Parenteral and Enteral Nutrition. Available at: Accessed July 31, 2014. 
  8. JCIA : Joint Commission International: accreditation and certification. Available at: Accessed July 2, 2014.
  9. Lang T, Secic M. How to report statistics in medicine 2nd edition; American College of Physicians, Philadelphia 2005: 137-8.
  10. Dawson B, Trapp R. Basic and Clinical Biostatistics 4th edition; McGraw-Hill 2004: 305-10.
  11. National Nutrition Survey 2008, Philippines. Available at: Accessed July 15, 2014.
  12. Lacuesta-Corro L, Paguia G, Lorenzo A, Navarette D, Llido LO. The results of the validation process of a Modified SGA (Subjective Global Assessment) Nutrition Assessment and Risk Level Tool designed by the Clinical Nutrition Service of St. Luke's Medical Center, a tertiary care hospital in the Philippines. PhilSPEN Online Journal of Parenteral and Enteral Nutrition. Available at: Accessed July 10, 2014.
  13. Ocampo RB , Camarse CM, Kadatuan Y, Torillo MR. Predicting Post-operative Complications Based on Surgical Nutritional Risk Level using the SNRAF in Colon Cancer Patients: A Chinese General Hospital & Medical Center Experience. PhilSPEN Online Journal of Parenteral and Enteral Nutrition. Available at: Accessed July 1, 2014.
  14. Young AM, Kidston S, Banks MD, Mudge AM, Isenring EA. Malnutrition screening tools: comparison against two validated nutrition assessment methods in older medical inpatients. Nutrition 2013; 29(1): 101-6.

Abstract | Introduction | Methodology | Results | Discussion | References | Back to Articles Page