Presented at the Neonatal Society 2015 Autumn Meeting.
Longford NT, Ramirez O, Giffe C
Imperial College, London
Universidad del Valle, Cali, Colombia
Background: Casa Canguro is a programme of health-care for infants with birth weight lower than 2.5 kg in Colombia. It includes about 5% of all births. The enrolled infants and their mothers are invited to visit appointed health-care clinics on prearranged dates. The body mass (weight), height and other basic anatomical and physiological measurements are recorded at each visit on paper forms. Of interest are the patterns of growth of infants in the first year of their lives within the groups defined by the gestational age at birth. The study aims to generate hypotheses about the growth of infants with low birth weight. Here we compare the pattern of growth of the studied infants up to their first birthday.
Methods: Weight and height of infants made at irregular time points up to the age of one year is analyzed. Over 30,000 visits were made by 2850 infants in 2007-2013. The data is cleaned in two rounds by specifying the largest plausible reversal in growth. The catch-up growth phenomenon is studied by comparing infants born at gestational ages 33-36 weeks and up to 31 weeks. The potential outcomes framework is applied by matching the two groups on an extensive list of background variables. Interpolation is used for weights at ages between two visits of an infant. The quality of the match is assessed balance plots. An extensive sensitivity analysis is conducted.
Results: Infants born at gestational age of 31 weeks or earlier first lag behind those born at 33 weeks or later, on average, by 0.57 kg at birth, 0.73 kg at 50 days of age and 0.89 kg at 91 days, but then gradually reduce the gap, so that the estimated difference at the age of one year is only 0.23 kg on average. Similar results are obtained for height; the difference at birth is 4.7 cm, and it gradually decreases to 1.3 cm at the first birthday. The standard errors for all the quoted estimates are so small that all the estimates are nominally highly significant. Sensitivity analysis, in which the settings of the original analysis are altered, confirms the robustness of these results. The settings studied include the number of propensity groups used in matching, the propensity model and related details, and the way how two infants in a matched pair are compared: by the difference of values, or whose measurement has a greater value.
Conclusion: The study of a register of births and visits of Colombian infants with low birth weight shows that the average weights of preterm and extreme preterm infants converge, on average, so that their differences are very small on average at their first birthdays. This is in accord with findings of convergence of growth of preterm and full-term infants at later ages.
Corresponding author: n.longford@imperial.ac.uk
Acknowledgements: Hospitality of Universidad del Valle, Cali, Colombia, accorded to the first author in November 2014, is acknowledged.