More than milk, eggs and orange juice / the author responds

Low birth weight, which Kramer calls "not a very useful outcome" (emphasis in the original) has been described as "one of the most important biologic predictors of infant death and deficiencies in physical and mental development during childhood among those babies who survive."2 The World Health Organization recommends the use of Williams' birthweight curve to diagnose small-for-- gestational-age births. The cut-off for a small-for-gestational-age term infant is 2900 g, and mounting evidence indicates that infants whose birth weight is above 2500 g but below about the lOth percentile still have higher health and nutrition risks than those whose birth weight is above the 10th percentile.2 In a 1987 meta-analysis3 Kramer concluded that in developed countries the most important risk factors for intrauterine growth retardation were low maternal energy intake or weight gain and low pregravid weight, and that low pre-pregnancy weight was the only important nutrition-related risk factor for preterm birth. A more recent study found that both preterm labour and small-for-gestational-age births were associated with several factors, including low pre-pregnancy weight and low weekly maternal weight gain.4 Other studies have found a relation between gestational weight gain and preterm delivery.5-7 In 2 of these,6,7 inadequate weight gain during the last half of pregnancy or the third trimester was associated with a higher risk of preterm birth. Kramer attacks the Canada Prenatal Nutrition Program (CPNP) because it is based on the US Women, Infants, and Children (WIC) program. Yet on the basis of evidence that WIC reduced low-birth-weight births by 25% and very-low-birthweight births by 44%, the US General Accounting Office concluded that WIC was a cost-effective program, resulting in savings of US$2.89 to US$3.50 for each federal dollar spent during the first 18 years of life.8 Proponents of the WIC program in the US cite evidence of effectiveness from comparisons of WIC participants and nonparticipants. But women who participate in WIC (or any other public health program, for that matter) are different from those who do not. Women who deliver very early, for example, will not have had the same time to enrol in WIC as those who deliver at term. Thus preterm birth can lead to nonparticipation, and an observational study may well put the cart before the horse by attributing to WIC the lower rate of preterm birth among participants. Participants also tend to be more commi