Potential for misclassification of infants' growth increments by using existing reference data

Area of Science:

• Pediatric Growth Monitoring
• Biostatistics
• Infant Development

Background:

• Accurate assessment of infant growth is crucial for identifying potential health issues.
• Existing reference data and statistical methods are used to evaluate infant weight and length gains.
• Curve-fitting methods are commonly applied to growth data for prediction and analysis.

Purpose of the Study:

• To compare observed variances in infant growth with predicted variances from reference curve-fitting methods.
• To quantify the underestimation of variance by curve-fitting techniques.
• To assess the impact of underestimation on the misclassification of infant growth patterns.

Main Methods:

• Analysis of monthly weight and length gains in 96 Peruvian infants.
• Application of reference curve-fitting methods to Peruvian infant data.
• Comparison of observed variances with predicted variances from curve-fitting and interpolation.

Main Results:

• Predicted variance estimates were significantly lower than observed variances from 2 to 12 months of age (P < 0.0001).
• Underestimation of total variance ranged from 59% to 94% due to ignoring random infant deviations.
• Approximately 24-67% of infants were misclassified as abnormal gainers in evaluated intervals.

Conclusions:

• Standard curve-fitting methods significantly underestimate growth variance in infants.
• This underestimation leads to a high rate of misclassification, potentially misidentifying normal growth as abnormal.
• Refined statistical approaches are needed to accurately assess infant growth trajectories and avoid misclassification.