Elsevier

The Lancet

Volume 360, Issue 9334, 31 August 2002, Pages 659-665
The Lancet

Articles
Unravelling the fetal origins hypothesis: is there really an inverse association between birthweight and subsequent blood pressure?

https://doi.org/10.1016/S0140-6736(02)09834-3Get rights and content

Summary

Background

The association between birthweight and subsequent blood pressure levels has been considered to provide some of the strongest, and most consistent, support for the “fetal origins” hypothesis of adult disease. It had been estimated that a 1 kg higher birthweight is typically associated with a 2–4 mm Hg lower systolic blood pressure.

Methods

55 studies that had reported regression coefficients of systolic blood pressure on birthweight (with 48 further studies that reported only the direction of this association), and seven such studies within twin pairs, were identified. Each study was weighted according to the inverse of the variance of the regression coefficient (ie, “statistical size”), and combined using a “fixed effects” approach.

Findings

Among the 55 studies that reported regression coefficients, there was a clear trend (p<0·0001) towards weaker associations in the larger studies: −1·9 mm Hg/kg in those with less than about 1000 participants; −1·5 mm Hg/kg with about 1000–3000 participants; and −0·6 mm Hg/kg with more than 3000 participants. By contrast with the inverse associations reported in 52 of these 55 studies, only 25 of the 48 studies that did not report regression coefficients found an inverse association (p<0·0001 for heterogeneity). Almost all of these regression coefficients had been adjusted for current weight (whereas few were adjusted for potential confounding factors), and removal of this adjustment in the larger studies reduced the estimated association to −0·4 mm Hg/kg. For studies within monozygotic twin pairs, the combined estimate was −0·6 mm Hg/kg with adjustment for current weight, and was also reduced without this adjustment.

Interpretation

Claims of a strong inverse association between birthweight and subsequent blood pressure may chiefly reflect the impact of random error, selective emphasis of particular results, and inappropriate adjustment for current weight and for confounding factors. These findings suggest that birthweight is of little relevance to blood pressure levels in later life.

Introduction

One of the original stimuli for the “fetal origins” hypothesis of adult disease was the observation that areas of Britain with the highest rates of neonatal mortality (and, by inference, of impaired fetal growth) early in the 20th century tended to have the highest rates of coronary heart disease later in the century.1 Subsequently, many retrospective studies have investigated associations of birthweight and of various other birth-related measures (such as placental to birthweight ratio, ponderal index, abdominal and head circumference) with vascular disease risk factors and disease in later life. Birthweight has been the most widely studied measure in such retrospective studies (chiefly due to its availability from existing records or personal recall), and the evidence for an association of adverse outcomes with lower birthweight is considered to be strongest for blood pressure.2, 3

Based on review of multivariate regression coefficients from 28 studies reported by March 1996, involving a total of 15 000 people, it was previously estimated that a 1 kg higher birthweight is typically associated with a 2–4 mm Hg lower systolic blood pressure.4 A recent update of that review,5 which included regression coefficients from an additional 27 studies, involving over 367 000 people, continued to suggest an inverse association of −2 mm Hg/kg (as did another recent review of the same studies6). But studies that had not reported the regression coefficient for this association did not contribute to those quantitative estimates, and no allowance was made for the size of the contributing studies. Moreover, whereas almost all of the available regression coefficients had been adjusted for measures of current weight when blood pressure was assessed, few involved adjustment for other potential confounding factors. The purpose of the present paper is to explore the possible impact of these issues, and so determine the likely relevance of birthweight to subsequent blood pressure.

Section snippets

Methods

Studies reporting by March 2000 on the association between birthweight and subsequent blood pressure had been identified previously for two systematic reviews of the available literature.4, 5 Details of the search strategies for such studies, and the inclusion and exclusion criteria, are provided in those reviews. There were 55 eligible studies (ie, individual cohorts, or subsets analysed separately) that had reported regression coefficients of systolic blood pressure on birthweight (web

Impact of publication bias on the apparent association

All but three of the 55 regression coefficients included in the previous reviews4, 5 reported an inverse association between birthweight and later blood pressure (table 1). But ordering the studies according to their statistical size (see Methods) yields a clear trend towards weaker associations in the larger studies (figure 1). For studies with statistical size less than 2 (typically involving fewer than 1000 participants) the inverse-variance-weighted estimate is 1·9 mm Hg lower systolic

Discussion

The present analyses indicate that bias in the reporting of results from studies of the association between birthweight and subsequent blood pressure may have led to substantial over-estimation of the strength of this apparent association. The larger studies are less likely to be prone to such bias, and consideration of the results from those that reported regression coefficients yields a weighted estimate of 0·6 mm Hg lower systolic blood pressure per 1 kg higher birthweight (compared with

References (35)

  • Breast cancer and hormone replacement therapy: collaborative reanalysis of individual data from 51 epidemiological studies including 52 705 women with breast cancer and 108 411 women without the disease.

    Lancet

    (1997)
  • K Christensen et al.

    Do genetic factors contribute to the association between birth weight and blood pressure?

    J Epidemiol Commun Health

    (2001)
  • K Dickersin

    How important is publication bias? A synthesis of available data.

    AIDS Educ Prev

    (1997)
  • KS Joseph et al.

    Review of the evidence on fetal and early childhood antecedents of adult chronic disease.

    Epidemiol Rev

    (1996)
  • N Paneth et al.

    Early origin of coronary heart disease (the “Barker hypothesis”): hypotheses, no matter how intriguing, need rigorous attempts at refutation.

    BMJ

    (1995)
  • CM Law et al.

    Initiation of hypertension in utero and its amplification throughout life.

    BMJ

    (1993)
  • MH Said et al.

    Relations between blood pressure at 3–4 years of age and body mass at birth: a population-based study.

    Rev Epidemiol Santé Publique

    (1998)
  • Cited by (0)

    View full text