SirMy thanks to Dr Kabir for pointing out an oversight in my recent paper on birthweight, namely that I did not deal with issues specific to developing countries.1 I am grateful for the opportunity to correct that omission.
The situation in the developing world is indeed different in key respects. Birthweight data are not necessarily free or plentiful. Gestational data may be non-existent. Mean birthweights can differ strikingly from those in Europe or the US. What do all these things imply for the study of birthweight?
The fundamental issues are no different. The distribution of birthweight has two components that are found in all populations. One is the predominant distribution of term births, and the other is a residual distribution of small pre-term births. The first determines the overall mean weight for births (although without indicating anything about the perinatal health of the population). The secondalbeit harder to estimateidentifies an important high-risk group of small babies.
These two components characterize every known distribution of human birthweights. Data may be harder to obtain in India or West Africa, but there is no reason to expect birthweight to behave differently there than elsewhere.
These two components of birthweight are independent. The importance of this fact cannot be overstated. We cannot predict one from the other. Dr Kabir says the mean birthweight in India is 2900 g compared with 3500 g in Sweden. This dramatic difference almost certainly reflects a difference in the two predominant distributions, which does not tell anything about either perinatal risk or residual births. A more crucial piece of information about the two countries would be the per cent of babies in their residual distributions. Dr Kabir surmises that the residual would necessarily be greater in India. It is impossible to predict that on the basis of the means. The residual may be higher in India, but that remains to be demonstrated. (Does anyone have data?)
The residual distribution may seem an abstract concept, but it corresponds to something concrete and practical. It is the proportion of babies that are both small and pre-term. Since it is a property of a distribution, it takes some statistical software to describe it precisely. Now that this software is operating on a website,2 the analysis is something you can carry out from the comfort of your nearest cybercafeas easily available these days in Kathmandu as in Cleveland.
Which brings me to Dr Kabirs point about the lack of gestational-age data in developing countries. This is one more reason why researchers in developing countries should welcome a method to extract information from birthweight distributions. The residual distribution is, after all, the only aspect of birthweight that sheds light on preterm births. Given that we know practically nothing about the risk of pre-term birth in developing countries, the analysis of residual distributions of birthweight would be a helpful (and fascinating) first step.
References
1 Wilcox AJ. On the importanceand the unimportanceof birthweight. Int J Epidemiol 2001;30:123341.
2 http://eb.niehs.nih.gov/bwt/