Comparison of Validated, Inpatient, Pediatric Nutritional Risk Screening Tools at
a Large, Urban, Tertiary Hospital
Abigail Olmstead¹, Elizabeth Spoede, MS, RD, CSP, LD², Elisabeth Hastings, MPH, RD, LD, CSSD², Molly Vega, MS, RD, LD²
¹Nutrition and Metabolism, University of Texas Medical Branch, ²Clinical Nutrition Services, Texas Children’s Hospital
Overview Methods Conclusion
• Texas Children’s Hospital’s current Nursing Nutrition Screening This was a cross-sectional study with data collection from October 9th, 2019 through February 14th, 2020 Summary of Findings:
Tool (NNST) has not been validated •STAMP most feasible based on reliability of questions
• Two validated pediatric screening tools were selected to pilot on •STAMP most sensitive (test positive risk and positive for
Screen Completion:
four floors of Texas Children’s Hospital at the Medical Center in •STAMP and STRONGKids criteria combined into one form, Nurse Admission Nutrition Risk malnutrition) based on RN screen and H&P note, STRONGKids very
Houston, TX Screen (Figure 1), questions edited from original screens for clarity. Questions 1 and 2 gathered from similar
• Results of screens were compared to the NNST and assessed for STAMP2, questions 3-6 gathered from STRONGKids3 •NNST greatest discrepancy of sensitivities between RN and H&P
feasibility and accuracy •Data collected within 24 hours of admission by RNs screen results
• RN completed screens were also compared to information •Additional demographic information obtained from electronic medical record (EMR) Epic Conclusions:
collected from the H&P note and assessed for reliability Data Entry: •The way the NNST is being completed currently is not as feasible or
• NNST showed the greatest discrepancy of sensitivities between •Demographic and screen data entered into a form created in REDCap (a HIPAA-compliant, secure
sensitive as the other two validated screens
RN completed screen and H&P data, STRONGKids and STAMP web application)
•Unclear which validated screen would be optimal for this population
•Z-Scores calculated using WHO and CDC growth charts for 0-36 months and 2-20 years,
were similar based on data
respectively
• In the future, work towards a two part screen, using BMI/weight- •Other hospital systems currently using the NNST should also
Statistical Analysis:
for-length z-scores <-1 as an automatic referral to a dietitian, then •Descriptive statistics created in REDCap for demographic data consider a reassessment of validity of the screen for their population
further assess using a validated pediatric screen •Feasibility of screens evaluated by comparison of nursing screen to intake (H&P) note in Epic for
inter-rater reliability (Table 2)
•Sensitivity evaluated by comparison of screen to clinically accepted malnutrition classification of
BMI/WFH z-score of <-1 and <-2 (Figure 3)
Project Aims •STAMP and STRONGKids scores measured as high or moderate risk recorded as positive.
•T-Tests calculated using Microsoft Excel
Future Directions
To evaluate the sensitivity and feasibility of two validated Total number of screens used: n=195 •P-value <0.05 considered significant • Conduct a focus group of RNs involved to further assess
pediatric nutritional risk screening tools, STAMP1 and Figure 2: Data Inclusion and Distribution feasibility, user satisfaction, and barriers of completion
STRONGKids2, in comparison to the currently utilized • Create and pilot a two part screen, utilizing a BMI/WFL z-score
Nursing Nutrition Screening Tool in a pediatric inpatient
setting.
Results
The group held a brainstorming session to identify the top opportunities for improvement, and proceeded with the following PDSA cycles of <-1 as an automatic referral, then assess further using
STRONGKids or STAMP
Table 1: Demographics Figure 3: Sensitivity of Screens • Standardize STRONGKids question 6 to improve feasibility
PDSA 1: The creation of the current state map • Implement training of screening tool for RNs to improve
NNST Questionnaire (positive screen if any yes): Total
West LegacyPDSA 2: Protocol Standardization reliability between screens and H&P note
1.0-36 months: Weight for length less than or equal to 10%? Tower Tower
PDSA 3: 7/1/16 – Education to birthing centers, hospitals, and pediatricians’ • Further studies to assess feasibility of new screen data entry
2.3-20 years: BMI for age less than or equal to 10% Sex, M:F (%) 45:55 39:61 56:44 using Epic
[Link] with Failure to Thrive? Age (y), Median 6.63 9.65 4.27
[Link] with Malnutrition? (range) (2d-17.9y) (2d-17.9y) (9d-17.3y)
[Link] with Anorexia Nervosa/Bulimia? Length of Stay (d), 3.58 2.49 5.72
[Link] with Adolescent Pregnancy? Median (range) (0.3-140.5) (0.3-53.7) (0.6-140.5) * *
[Link] with other Nutrition Risk? (specify) Diagnosis (%)
Respiratory: 39 32.6 52.4
Gastrointestinal: 4.1 4.5 3.2
Endocrine: 12.8 18.9 0
2.1 1.5 3.2
Figure 1: Nurse Admission Nutrition Risk Screen Cardiac:
Renal: 2.1 2.3 1.6 References
Neurological: 0.5 0 1.6
Oncologic: 1 1.5 0 WFH: Weight for height
* Statistically Significant difference (p-value
1. McCarthy, H., Dixon, M., Crabtree, I., Eaton ‐Evans, M.J. and McNulty, H.; The
Malnourishment: 3.6 5.3 0 development and evaluation of the Screening Tool for the Assessment of
12.8 9.8 19 <0.05) Table 2: Feasibility of Screens
Surgical: Malnutrition in Paediatrics (STAMP©) for use by healthcare staff. Journal of Human
Rheumatic: 0.5 0.8 0 Nutrition and Dietetics. 2012; 25: 311-318.
Comparison of Data from Screen vs H&P P-Value
Infectious: 4.6 6.1 1.6 2. Jessie M. Hulst, Henrike Zwart, Wim C. Hop, Koen F.M. Joosten; Dutch national
1. Metabolic: 0 0 0 NNST 4.13E-05 ** survey to test the STRONGkids nutritional risk screening tool in hospitalized
Other: 16.9 17 17.5 children; Clinical Nutrition. 2010; 29:106-111.
STAMP 0.27
2.
Weight for age Z- -0.14 (1.58) -0.11 (1.52) -0.21 (1.71) STRONGKids 3.23E-05 **
3.
Score, Mean (SD)
** p-value of <0.001 Acknowledgements
4. Length for age Z- -0.54 (1.74) -0.46 (1.53) -0.72 (2.13)
Comparison of similar questions between STAMP and STRONGKids:
5.
Score, Mean (SD)
• No significance in admitting diagnosis (questions 1 and 6)
We would like to thank the nursing staff of Texas Children’s
BMI Z-Score, Mean 0.27 (1.77) 0.21 (1.54) 0.39 (2.18) • Significant difference (p-value <0.001) in current intake/symptoms Hospital for completing the screens and gathering
6. (SD)
(questions 2 and 3) anthropometric data.
© Texas Children’s Hospital. All rights reserved.
