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

Biostatistics Unit, Centre for Epidemiology and Biostatistics, University of Leeds, Leeds LS2 9LN, UK.
American Journal of Epidemiology (Impact Factor: 5.23). 02/2005; 161(1):27-32. DOI: 10.1093/aje/kwi002
Source: PubMed


Some researchers have recently questioned the validity of associations between birth weight and health in later life. They argue that these associations might be due in part to inappropriate statistical adjustment for variables on the causal pathway (such as current body size), which creates an artifactual statistical effect known as the "reversal paradox." Computer simulations were conducted for three hypothetical relations between birth weight and adult blood pressure. The authors examined the effect of statistically adjusting for different correlations between current weight and birth weight and between current weight and adult blood pressure to assess their impact on associations between birth weight and blood pressure. When there was no genuine relation between birth weight and blood pressure, adjustment for current weight created an inverse association whose size depended on the magnitude of the positive correlations between current weight and birth weight and between current weight and blood pressure. When there was a genuine inverse relation between birth weight and blood pressure, the association was exaggerated following adjustment for current weight, whereas a positive relation between birth weight and blood pressure could be reversed after adjusting for current weight. Thus, researchers must consider the reversal paradox when adjusting for variables that lie within causal pathways.

Download full-text


Available from: Yu-Kang Tu, Nov 08, 2015
  • Source
    • "Thus, evidence from literature suggests limited possibility of bias in current study due to confounding. Furthermore, adjustment for adult body size for variables, which are on the causal pathway between birth weight/childhood body size and MD/breast cancer risk, is arguably inappropriate and may lead to an artifactual statistical effect [44]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: High breast density, a strong predictor of breast cancer risk, may be determined early in life. Childhood anthropometric factors have been related to breast cancer and breast density, but rarely simultaneously. We examine whether mammographic density (MD) mediates an association of birth weight, childhood body mass index (BMI) and height with the risk of breast cancer. A total of 13,572 women (50-69 years) in the Copenhagen mammography screening program (1991-2001) with childhood anthropometric measurements in the Copenhagen School Health Records Register were followed for breast cancer until 2010. Using logistic and Cox regression models we investigated associations among birth weight, height and BMI at ages 7-13 with MD (mixed/dense or fatty) and breast cancer, respectively. A total of 8,194 (60.4%) women had mixed/dense breasts and 716 (5.3%) developed breast cancer. Childhood BMI was significantly and inversely related to having mixed/dense breasts at all ages, with age at screening and birth cohort adjusted odds ratios (95% confidence intervals) ranging from 0.69 (0.66-0.72) at age 7 to 0.56 (0.53-0.58) at age 13, per one unit increase in z-score. No statistically significant associations were detected between birth weight and MD, height and MD, or birth weight and breast cancer risk. BMI was inversely associated to breast cancer risk, with age and birth cohort adjusted hazard ratios (HRs) of 0.91 (0.83-0.99) at age 7 and 0.92 (0.84-1.00) at age 13, whereas height was positively associated with breast cancer risk (age 7: 1.06 (0.98-1.14) and age 13: 1.08 (1.00-1.16)). After additional adjustment for MD, associations of BMI with breast cancer risk diminished (age 7: 0.97 (0.88-1.06) and age 13: 1.01 (0.93-1.11)), but remained with height (age 7: 1.06 (0.99-1.15) and age 13: 1.09 (1.01-1.17)). Among women 50 years and older, childhood body fatness was inversely associated to the breast cancer risk, possibly via a mechanism mediated by MD, at least in part. Childhood tallness was positively associated with breast cancer risk, seemingly via a pathway independent of MD. Birth weight was not associated with MD or breast cancer risk in this age group.
    Breast cancer research: BCR 01/2014; 16(1):R4. DOI:10.1186/bcr3596 · 5.49 Impact Factor
  • Source
    • "We also observed that the associations between skinfold thickness at birth and infancy visceral depth strengthened with further adjustment for current skinfold thickness. Some investigators have argued that adjustment for current size could potentially introduce bias due to overcontrolling [41]. However, such adjustment can be justified if current body size is a potential confounder that is positively associated with both birth size and the outcome of interest. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Other imaging techniques to quantify internal-abdominal adiposity (IA-AT) and subcutaneous-abdominal adiposity (SCA-AT) are frequently impractical in infants. The aim of this study was twofold: (a) to validate ultrasound (US) visceral and subcutaneous-abdominal depths in assessing IA-AT and SCA-AT from MRI as the reference method in infants and (b) to analyze the association between US abdominal adiposity and anthropometric measures at ages 3 months and 12 months. Twenty-two infants underwent MRI and US measures of abdominal adiposity. Abdominal US parameters and anthropometric variables were assessed in the Cambridge Baby Growth Study (CBGS), n = 487 infants (23 girls) at age 3 months and n = 495 infants (237 girls) at 12 months. US visceral and subcutaneous-abdominal depths correlated with MRI quantified IA-AT (r = 0.48, P < 0.05) and SCA-AT (r = 0.71, P < 0.001) volumes, respectively. In CBGS, mean US-visceral depths increased by ~20 % between ages 3 and 12 months (P < 0.0001) and at both ages were lower in infants breast-fed at 3 months than in other infants. US-visceral depths at both 3 and 12 months were inversely related to skinfold thickness at birth (P = 0.03 and P = 0.009 at 3 and 12 months, resp.; adjusted for current skinfold thickness). In contrast, US-subcutaneous-abdominal depth at 3 months was positively related to skinfold thickness at birth (P = 0.004). US measures can rank infants with higher or lower IA-AT and SCA-AT. Contrasting patterns of association with visceral and subcutaneous-abdominal adiposities indicate that they may be differentially regulated in infancy.
    Journal of obesity 04/2013; 2013(5):951954. DOI:10.1155/2013/951954
  • Source
    • "Responds to that critical view accentuated that the effect of birth weight is conditioned by events occurring later in life, like weight history [18] [19]. Furthermore, Tu et al. had questioned the use of current weight as a confounder, but suggested it as a covariate and additionally claimed that the evidence for the association between birth weight and hypertension is a statistical artifact [4]. was to examine associations between birth weight and adult hypertension, while considering the weight history between birth and adulthood, the familial predisposition to obesity and several important covariates known to influence the association between birth weight and hypertension in adulthood. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The fetal origin hypothesis suggests that birth weight is related to several adult diseases. One hypothesis is that low birth weight is associated with increased risk of hypertension. In the pre-sent study we analyzed the association between self-reported birth weight and risk of hyperten-sion with detailed confounder control, and the modification on this association by familial obe-sity and own weight history during childhood and youth. Methods: In 1999 the Danish Nurse Cohort study included 31,642 females aged 44 years or more of whom 76% replied to questions on birth weight, weight history, familial obesity disposition, parents socioeconomic status and presence of hypertension. Results: In total 14% reported birth weights below 3000 g. The uni-variate hazard ratio of hypertension for those born with birth weight < 3000 g was 1.24 (95% CI (1.13 - 1.36)) compared to those with birth weight > 3000 g. The association remained stable after adjustment for important confounders, as life-style, weight history and socioeconomic status during upbringing. Conclusion: Our results sup-port the hypothesis that low birth weight is as-sociated with hypertension in adult life, also after considering several confounding factors and weight history later in life. Furthermore, the previous weight history might have an influence on the association between birth weight and the risk for adult hypertension.
    Health 01/2013; 5(4A):16-22. DOI:10.4236/health.2013.54A003 · 0.51 Impact Factor
Show more