First, the good news. Neither of my excellent colleagues has defended the study of low birthweight (LBW).1,2 Hooray. Let us admit LBW is a worn-out approach, and put it away. I am also glad that neither commentator has defended the practice of ordinary adjustment for birthweight in the analysis of infant mortality. The essential assumptions that support such adjustment are nearly always violated. If everyone could agree on these two points, we would spare ourselves a lot of unproductive work.
So what is the bad news? I suppose it is the general confusion over where we go from here. The framework I propose3 has not been warmly welcomed by the commentators. Dick David implies that my eyes have slipped from the prizethe prize being the reduction of infant mortality. Using race as an example, he suggests that my attempts to steer attention away from birthweight as the source of racial differences in infant mortality are naïve and doomed to misinterpretation. Well, he got that right.
Ian Russell and I published an analysis of Black-White differences in the US4 which led us to conclude that a high rate of preterm deliverynot lighter mean weightis the major culprit in the high mortality of African-American babies. Efforts to raise average birthweights (through nutritional programmes, etc.) are unlikely to improve mortality. If we are going for the prize (namely, closing the race gap in mortality), then we should start with the excess of Black preterm births. Racism is likely to be an ingredient in this excess. Data from Collins et al. suggest that racial discrimination increases a mother's risk for small preterm births.5 But much of the US public health community remains committed to birthweight intervention. So whose eyes are on the prize?
Dick David says mean birthweights can be increasedmy question is whether they need to be. Infant mortality rates have fallen dramatically in recent decades with little or no increase in mean weights. Why couldnt African-Americans be as healthy at birth as European-Americans, regardless of their weight? Dick suggests I am politically incorrect but scientifically right. That's better than the other way around.
Regarding Irva Hertz-Picciotto's comments, I was not clear whether she wants to fine-tune my ideas or toss them out. I will try to address her particular points, saving the big one for last.
Irva says I lose sight of the distinction between identical mortality curves and parallel mortality curves (all this after adjusting for shifts in birthweight). MacMahon6 did not distinguish between those twobut I do. It is in fact at the foundation of my proposed approach to birthweight.
Another issue Irva raises is whether the adjusted mortality curves are always parallel.
Irva thinks she sees evidence of interaction with birthweight in one of the Figures. She may be right. I have not paid much attention to possible interactions in the smaller data sets because there is so little evidence for it in the huge national data sets. I would be happy to discuss these fine points of adjusted mortality curvesespecially if it means we all accept the premises that get us to that point.
In the same spirit, I am open to Irva's suggestion that the relation of birthweight to morbidity may sometimes differ from its relation to mortality. This is a hard question to address because there is not much information on birthweight-specific morbidity. So far, the available data suggest the patterns for weight-specific morbidity and mortality are the same.3
Irva's final point is that if birthweight is on the causal pathway, its use as a surrogate outcome is appropriate and informative. This is the crux of the matter. I do not think birthweight is on the causal pathway to mortalityat least not for the variables epidemiologists study.
This does not mean that birthweight is an uninteresting endpoint. If racial discrimination were shown to reduce birthweights, it would provide a compelling demonstration of the tangible effects of racism on physical processes. But this would not prove birthweight is the biological mechanism by which racism contributes to infant mortality.
Birthweight is a fascinating variable. Still, when it comes to the study of infant mortality, birthweight is a side issue. Richard David may be right when he says this is politically discomfiting. But if our conventional ways of thinking about birthweight and mortality are unsuccessful, then some newand uncomfortable options may be exactly what are needed.
References
1
David R. Commentary: Birthweights and Bell Curves. Int J Epidemiol 2001;30:124143.
2
Hertz-Picciotto I. Commentary: When brilliant insights lead astray. Int J Epidemiol 2001;30:124344.
3
Wilcox AJ. On the importanceand unimportanceof birthweight. Int J Epidemiol 2001;30:123341.
4 Wilcox AJ, Russell IT. Why small black infants have lower mortality than small white infants: the case for population-specific standards for birthweight. J Pediatr 1990;116:710.[ISI][Medline]
5 Collins JW Jr, David RJ, Symons R, Handler A, Wall SN, Dwyer L. Low income African-American mothers' perception of exposure to racial discrimination and infant birth weight. Epidemiology 2000;11:33739.[ISI][Medline]
6 MacMahon B, Alpert M, Salber EJ. Infant weight and parental smoking habits. Am J Epidemiol 1966;82:24761.[ISI]