[Variability of percentile values of body measurements of newborn infants (with special reference to body weight)].

The percentile curves for fetal growth between the 25th and 42nd week of gestation calculated for more than 52,000 newborns in East Germany in 1985 were compared with the rates defined for the same population in 1973-74 and showed a significantly lower mean birth weight. Besides methodological differences (primarily a more precise estimation of gestational age in the new study by using ultrasound fetometry), the differences seem to be mainly the result of changes in the reproductive behaviour of the population over that period (decrease in parity). When comparing different populations independently of ethnic specificities, a number of modifying factors have to be taken into account which significantly influence the birth weight, namely: parity, spacing, age and stature of the mother. These variables have also to be remembered when assessing the maturity of newborns in order to standardize the birth weight in relation to mortality and morbidity. As a basis for calculating percentile curves the cumulative frequency distribution should be preferred, rather than assuming a normal distribution, since fetal diseases or developmental anomalies result in an uneven distribution with an over-representation of small for date fetuses. The selection of data for calculating normal values must be rejected because of the arbitrary decisions which this entails.