A study of validity and reliability for Subjective Global Nutritional Assessment in outpatient children with cerebral...

To investigate the reproducibility, stability, internal consistency and the ability to grade malnutrition of Subjective Global Nutritional Assessment (SGNA) in outpatient children with cerebral palsy.

This was a part of a larger, cross-sectional study (ChiCTR2000033869) at the outpatient of a tertiary hospital. The recruitment and data collection of children with Cerebral Palsy aged from 1 to 18 years were from August 2020 to March 2021. The concurrent validity, inter-rater reliability, test-retest reliability and internal consistency of SGNA were tested. To analyze data, specificity, sensitivity, Kendall coefficient, Cohen's kappa coefficient, Spearman coefficient and Cronbach's α coefficient were used.

The agreement between SGNA and anthropometric data was moderate to strong (k = 0.540-0.821). The sensitivity (71.70% to 89.74%) and specificity (77.67% to 91.03%) of SGNA to identify participants with z-score ≤-2 were good. The sensitivity of SGNA to identify participants with weight for age z-score ≤-3 was poor (30.00%). The interrater reliability (k = 0.703) and test-retest reliability (k = 0.779) were good. The item of edema was with poor agreement to SGNA nutritional grades (r_s = 0.072), and after deleting it from SGNA, the Cronbach's α coefficient of SGNA increased from 0.736 to 0.871.

SGNA is good at identifying malnourished outpatient children with cerebral palsy, with excellent reproducibility and short-time stability. However, the ability to grade malnutrition is unsatisfactory. For further application in this group, a more appropriate item should be designed to replace the item of edema.