We appreciate many of the points raised in Dr. Gillman's letter (1) on our article (2) and our response (3) to Dr. Weinberg (4). However, these points are not strictly pertinent to the key message we sought to convey. We demonstrated that adjusting for current body weight can induce a spurious inverse statistical association between birth weight and health in later life. We accept that the model we used was simplistic compared with the more complex models proposed by Weinberg (4), but this does not mean that our conclusions were not valid. Our key message was that statistical models should be based on the likely biologic mechanisms of the hypothesis under test if the models' findings are to be interpreted as a test of the mechanisms inferred. Considering the vast literature on the fetal origins hypothesis, we question why so few studies have justified their statistical adjustment for one or more measures of current body size by formulating clear causal models and associated research hypotheses in which adjusting for current body size is necessary or appropriate.

We recognize that birth weight is generally treated as a proxy for fetal growth in fetal origins analyses, but we disagree that this means that “trying to estimate the overall ‘effect’ of birth weight on blood pressure is nonsensical” (1, p. 292). On the contrary, we argue that estimating the independent effect of birth weight on blood pressure by adjusting for measurements of current body size is itself nonsensical when “higher birth weight is related to higher body mass index, and higher body mass index predicts higher blood pressure” (1, p. 292). Our analyses clearly demonstrate that it is not always feasible to estimate the isolated or “independent” effects of birth weight using multiple regression analysis (2, 3), and attempting to do so is likely to be misleading (5). Furthermore, we wonder whether all proponents of the fetal origins hypothesis would agree with the statement that “in the ‘fetal origins’ paradigm… birth weight itself is not a causal factor” (1, p. 292). It is clear from the literature on the fetal origins hypothesis that size at birth is considered an important marker for prenatal development and/or nutrition and that small size at birth is taken to be an indication of restricted development in utero due to prematurity, undernutrition, placental insufficiency, and/or other developmental problems. However, most studies on the fetal origins hypothesis do not focus on the beneficial effects on adult health of preventing the restricted growth or prematurity of babies with extremely low birth weight but instead focus on babies within the normal range of birth weight. Therefore, while we agree with Dr. Gillman (1) that birth weight itself may not be a causal factor, we would ask 1) what birth weight might be a proxy for; 2) why babies with a slightly smaller than average birth size are considered to have restricted development in utero; and, especially, 3) why it is necessary or appropriate to adjust for current body size to test the fetal origins hypothesis. We welcome more experimental studies to directly determine the causal role (if any) of putative prenatal factors directly, but we do not understand why Dr. Weinberg's suggestion of investigating twins (4) is considered unhelpful.

Finally, it is interesting to note that Dr. Gillman's letter (1) refers to the fetal origins hypothesis as a “paradigm”—a term made popular by Thomas Kuhn (6). When a scientific paradigm is challenged by anomalies that contradict the predictions it makes, proponents of the paradigm can be tempted to ignore such anomalies or propose ad hoc rules which exclude the anomalies from the paradigm. We hope the fetal origins paradigm does not suffer either fate but can accommodate the issues we have raised.

Conflict of interest: none declared.

References

1.
Gillman MW. Re: “Why evidence for the fetal origins of adult disease might be a statistical artifact: the ‘reversal paradox’ for the relation between birth weight and blood pressure in later life.” (Letter).
Am J Epidemiol
 
2005
;
162
:
292
.
2.
Tu YK, West R, Ellison GTH, et al. Why evidence for the fetal origins of adult disease might be a statistical artifact: the “reversal paradox” for the relation between birth weight and blood pressure in later life.
Am J Epidemiol
 
2005
;
161
:
27
–32.
3.
Tu YK, Ellison GTH, West R, et al. Tu et al. respond to “Barker meets Simpson.”
Am J Epidemiol
 
2005
;
161
:
36
–7.
4.
Weinberg C. Invited commentary: Barker meets Simpson.
Am J Epidemiol
 
2005
;
161
:
33
–5.
5.
Kaufman JS, MacLehose RF, Kaufman S. A further critique of the analytic strategy of adjusting for covariates to identify biologic mediation.
Epidemiol Perspect Innov
 
2004
;
1
:
4
. (Electronic article). (World Wide Web URL: http://www.epi-perspectives.com/content/1/1/4).
6.
Kuhn T. The structure of scientific revolution. 2nd ed. Chicago, IL: University of Chicago Press,
1970
.