Objective: To evaluate the performance and the utility of using birthweight-adjusted scores of the Dubowitz method of estimating gestational age in a Zimbabwean population.
Design: A validation study.
Setting: Harare Maternity Hospital, from October to December 1999.
Subjects: 364 African newborn infants with a known last menstrual period (LMP), within the first 56 hours of life.
Main outcome measures: Differences between regression lines and variances explained by Dubowitz scores obtained by examining newborn infants compared to gestational age calculated from the last menstrual period, in models with and without the addition of birthweight.
Results: The Dubowitz method was a good predictor of gestational age, useful in differentiating term from pre-term infants. The beta coefficients from regression lines with and without addition of birthweight differed significantly from each other (z = 2.83, p < 0.01). Our regression line without adding birthweight was Y(LMP gestational age) = 23.814 + 0.301* score. Addition of birthweight to the regression models improved prediction of gestational age, Y(LMP gestational age) = 23.512 + 0.219* score + 0.0015* grams, and accounted for 69% of the variance compared to 66% in models without birthweight.
Conclusion: The introduction of birthweight improves estimation of gestational age, correcting for the overestimation reported for the original Dubowitz methods and the error caused by low birthweight. We recommend the use of our birthweight-adjusted Dubowitz maturity scales for studies of prematurity, and for routine clinical practice.