Objective: Birth weight percentiles based on weekly measurements are used to assess the nutritional status of preterm infants. However, as preterm infants exhibit a rapid growth rate (up to 20 g/kg/day), their body weight can increase by 15% per week. We calculated birth weight percentiles based on daily measurements, to more precisely classify very preterm infants (gestational age of 154-223 days).
Methods: Data of 23,864 (10,720 females and 13,144 males) very preterm singleton infants with a gestational age of 154-223 days (22-31 completed weeks) were retrieved from the German perinatal statistics of 1995-2000. Percentile curves based on the empirical birth weight data were subjected to three statistical smoothing procedures: cubic regression, local regression (LOESS smoothing), and the LMS method.
Results: Smoothing of the birth weight percentiles using cubic regression produced the smallest residual variance.
Conclusion: Birth weight percentiles based on daily averages allow a more precise assessment of the somatic development of preterm infants.