Rapid Responses to:

PAPERS: Terence Dwyer, Leigh Blizzard, Ruth Morley, and Anne-Louise Ponsonby Within pair association between birth weight and blood pressure at age 8 in twins from a cohort study BMJ 1999; 319: 1325-1329 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Within pair association between birth weight and blood pressure at age 8 in twins

Collings   (26 November 1999)

Authors' reply

T Dwyer, L Blizzard, R Morley, A L Ponsonby   (13 December 1999)

Syndrome X and the Fetal Origins Hypothesis

Patrick J Bradley   (19 December 1999)

Early Growth Programs Later Blood Pressure in Twins Born Preterm

Atul Singhal   (11 August 2000)

Why not report the result about diastolic and mean blood pressure?

Clayman ZK Zhang   (19 April 2002)

Within pair association between birth weight and blood pressure at age 8 in twins 26 November 1999
Collings , Medical students Dept of Epidemiology & Public Health, University of Newcastle Send response to journal: Re: Within pair association between birth weight and blood pressure at age 8 in twins	Dear Editor The use of multiplets provided Dwyer et al1 with an original way to test the Barker Hypothesis. This allowed the effect of genetic and maternal influences on childhood blood pressure to be analysed separately. Although we feel that the study largely supported the authors conclusions, there are a few points we feel are worthy of consideration. Firstly, we note that all the singletons in this study were originally selected because of their possible susceptibility to SIDS. Was this a truly representative sample, suitable for generalisation to a larger population ? Secondly, no mention is made of any confounding factors which would have been in effect during the post natal period. Table 1 shows large variations in mean birth weights between boy and girl singletons and twins although their current weight is very similar. Therefore, could the rate of weight gain rather than birth weight be the more important variable? Another possible confounding variable may be that low birth weight children are fed differently. Higher calorie or salt content feeds could also contribute to the increase in systolic BP seen in the low birth weight groups. Finally, the authors placed a heavy emphasis on the use of within pair analysis, confidently reporting a negative association between birth weight and blood pressure. However, the confidence intervals used for this analysis are large and could actually demonstrate a positive association in all three groups. We support the authors' conclusion that further research into the functioning of the fetoplacental unit is necessary. Yours sincerely A Collings E Sconce J Wylie V Blaylock K Morgan Third Year Medical Students Department of Epidemiology & Public Health, The Medical School, University of Newcastle Word Count 250 words Reference: (1) Dwyer T, Blizzard L, Morley R, Ponsonby A.L. Within pair association between birth weight and blood pressure at age 8 in twins from a cohort study. BMJ.1999;319:1325-1329.
Authors' reply 13 December 1999
T Dwyer , L Blizzard, R Morley, A L Ponsonby Send response to journal: Re: Authors' reply	Firstly, Collings et al question whether our results are generalisable because the singletons in our study were a non- representative sample of their birth cohort. The inferences we drew in the paper, however, were based mainly on the analysis of blood pressure and birth weight in multiplets (mainly twins). All multiplets born in Tasmania were eligible for inclusion in the Tasmanian Infant Health Survey (TIHS). The sample we studied at 8 years comprised 85% of all multiplets born in 1988 or 1989 and living in southern Tasmania in 1996-97. Collings et al raise an interesting point about external validity. The singletons were a 1-in-5 sample of live births in Tasmania during 1988-89 and those with risk factors for SIDS, including male sex and low birth weight, are over-represented. Does it matter if a sample such as this is not representative of the population? If the purpose of the study is to test hypotheses about a causal association within a biological framework, the answer is no. What does matter (3) is that the sample contains a wide distribution of the study factor and its effect modifiers. Miettenen (3) reminds us that the empirical relation of body weight to gender does not depend on the gender distribution in the study base, but we need sufficient subjects of each gender in the sample to estimate it with acceptable precision. Studies are also ideally (3) and often designed to limit the distribution of key confounders, by restricting the range of ages of subjects in the sample as one example, but it is not always possible to be certain in advance whether a factor will prove to be a confounder or an effect modifier (4). There is no a priori reason for assuming that the optimal distribution for assessing effect modification and controlling confounders is found in a representative sample of some population. Secondly, Collings et al suggest that our results are biased by the failure to account for post-natal confounding factors. They cite post- natal weight gain and diet as examples of such factors. In fact we did adjust for current size (fatness at age 8 years) when examining the birth weight- blood pressure association. This precluded us from separately controlling for growth to age 8 because of the exact linear dependence (childhood weight = birth weight + weight gain). This left a choice of which two of these variables to include in the model. We chose birth weight because it is the most immediate outcome of the fetoplacental environment, and we chose current fatness because of the involvement of upper body weight and obesity in hypertension, glucose intolerance and hypertriglyceridemia. We recognise, as have others (5), that the estimated effects of birth weight and current weight are biased if growth is also a causal factor. We did, nonetheless, look to see whether growth during the first month of life, measured in grams or as a percentage of birth weight, was more strongly associated with childhood blood pressure that birth weight, but found it was not. In relation to diet, we do not have the data on infant feeding and nutrition to test the hypotheses suggested, but consider it unlikely that this could explain the differences found within twin pairs. Thirdly, Collings et al question the way we interpret the within pair association between birth weight and blood pressure. We believe it was important to show how twins could be used to obtain insights into causal pathways that might explain the 'fetal origins' hypothesis. They are correct in stating that our data did not conclusively demonstrate that within pair analysis failed to remove the negative birth weight-blood pressure association. The confidence intervals were provided to show that, if this study of just 55 twin pairs was repeated many times, positive associations could be expected on some occasions (the model estimate is 10.5% of repetitions). Equally, negative values larger in absolute size than our estimate (-5.3 mm per kilogram) would occur on 50% of occasions. We undertook the study just once, and our best estimate based on this one sample is that a reduction in blood pressure of 5.3 mm per kilogram increase in birth weight remained after pairing. Overall, our data were more consistent with the inference that an effect remained than with the alternative that it was removed. Our final comment that the findings needed to be replicated in other studies reflected our own concerns that chance alone could explain the lack of change in the effect estimate after pairing. The confirmation of our results by the accompanying paper (6) increases the likelihood that the inference was correct. 1. Collings A, Sconce E, Wylie J, Blaycock V, Morgan K. Re: Within pair association between birth weight and blood pressure at age 8 in twins from a cohort study.[Electronic response] BMJ 26 November 1999. 2. Dwyer T, Blizzard T, Morley R, Ponsonby A-L. Within pair association between birth weight and blood pressure at age 8 in twins from a cohort study. BMJ 1999;319:1-5. 3. Miettenen O. Theoretical Epidemiology: principles of occurrence research in medicine. New York: Wiley 1985:46-58.. 4. Ponsonby A-L, Dwyer T, Couper D. Is this finding relevant? Generalisation and epidemiology. Aust NZ J Public Health 1996;20:54-6. 5. Lucas A, Fewtrell MS, Cole TJ. Fetal origins of adult disease-the hypothesis revisited. BMJ 1999 Jul 24;319(7204):245-9. 6. Poulter NR, Chang CL, MacGregor AJ, Snieder H, Spector TD. Association between birth weight and adult blood pressure in twins: historical cohort study. BMJ 1999;319:6-9. Terence Dwyer Director Leigh Blizzard Research Fellow (Biostatistician) Menzies Centre for Population Health Research, University of Tasmania, GPO Box 252-23, Hobart TAS 7001, Australia Ruth Morley Senior research fellow Clinical Epidemiology and Biostatistics Unit, University of Melbourne Department of Paediatrics, Royal Children's Hospital, Melbourne, Victoria, Australia Anne-Louise Ponsonby Public health physician Academic Unit of General Practice and Community Care, Canberra Clinical School, University of Sydney, Canberra, Australian Capital Territory, Australia
Syndrome X and the Fetal Origins Hypothesis 19 December 1999
Patrick J Bradley, General Practitioner Berry Street Medical Centre, 14/199 Pacific Hwy, North Sydney NSW 2060, AUSTRALIA Send response to journal: Re: Syndrome X and the Fetal Origins Hypothesis	Dear Editor It may be no coincidence that the conditions associated with Syndrome X – hypertension, non-insulin-dependent diabetes (NIDD) and coronary heart disease (CHD) – are also associated with the “fetal origins hypothesis”. The dietary factors linked to Syndrome X include increased calorie intake, increased fat intake and possibly refined sugars. Thus it is possible that low weight new-borns consume more calories and fat during their catch-up growth and it is this that predisposes them to future NIDD and CHD. The increased strength of the association between birth weight and systolic blood pressure in twins compared to singletons found by Dwyer et al (1) may reflect the fact that twins weigh 10% less than singletons at birth. Thus they have greater catch growth than singletons and this would amplify any effect of differences in nutrition between pairs of twins. The greater strength of the association between monozygotic twins compared to dizygotic twins may be because monozygotic twins not only weigh less than singleton newborns but are also less concordant for weight at birth than dizygotic twins (2) and this combination may also be an amplifying factor. Low birth weight per se does not appear to be a risk factor. Birth weight in developing countries has been much lower than in developed countries and this has been associated with an almost total absence of CHD and NIDD. It is only when developing countries Westernize their diet does CHD and NIDD appear and then at rates greatly in excess of those in developed countries. Thus if there is any association between birth weight and these disorders it would appear that heaviness in infancy is protective against the dietary factors causing Syndrome X. However even this may be an artifact as underweight infants not only consume more calories to catch up in weight but they may also often consume more calories per kilogram of body weight to maintain thermoregulation (3,4,5). Thus the “fetal origins hypothesis” may simply be another manifestation of that ubiquitous nutritional disorder – Syndrome X. Patrick J Bradley P.O. Box 765, Chatswood NSW 2057, AUSTRALIA 1. Dwyer T, Blizzard L, Morley R, Ponsonby A-L. Within pair association between birth weight and blood pressure at age 8 in twins from a cohort study. BMJ 1999; 319:1325-1329. 2. Wilson RS. Concordance in physical growth for monozygotic and dizygotic twins. Ann Hum Biol 1976;3:1-10. 3. Pittet PG. Direct calorimeter with fast response time using the gradient-layer principle: some illustrations of its utilization in human studies. In: Bjorntorp P, Cairella M, Howard AN, eds. Recent Advances in Obesity Research: III. London: John Libbey & Company Limited, 1981:146 -152. 4. Dubois S, Hill DE, Beaton GH. An examination of factors believed to be associated with infantile obesity. Am J Clin Nutr 1979;32:1997-2004. 5. Vobecky JS, Vobecky J, Shapcott D, Demers PP. Nutrient intake patterns and nutritional status with regards to relative weight in early infancy. Am J Clin Nutr 1983; 38:730-738.
Early Growth Programs Later Blood Pressure in Twins Born Preterm 11 August 2000
Atul Singhal MRC Childhood Nutrition Research Centre Send response to journal: Re: Early Growth Programs Later Blood Pressure in Twins Born Preterm	Sir, A twin study, published in the BMJ1, showed low birthweight was associated with higher blood pressure in childhood, independent of possible maternal and social confounding factors. We hypothesised that in subjects born preterm, more rapid ex-utero growth prior to full term might ‘programme’ lower blood pressure. To test this, we studied twins included in a cohort of 926 (<1850g birthweight), participating in randomised outcome trials2 of early diet and later cognitive function and cardiovascular risk2. Survivors (mean gestation: 30.4, SD: 2.5 weeks) were studied at age 8 (n=782) and a subset at 13-16 years (n=216), including 50 and 14 twin pairs respectively. Zygosity was unknown. Systolic and diastolic blood pressures were measured by an automated device (AccutorDatascope) for most subjects at age 82 and for all at 13-16 years. Mean blood pressure (area under blood pressure/time curve) at 13-16 years of the second-born twin was subtracted from that of the first-born. After adjusting for sex difference, between- pair difference in blood pressure was related to between-pair difference in birthweight1 or post-natal growth. A greater between-pair difference in weight gain to 2 weeks post- natally was inversely related to a greater between-pair difference at 13- 16 (but not 8) years, in diastolic pressure (regression coefficient (B) = -10.6 mm Hg/100g weight gain difference; 95% Confidence Interval: CI: - 16.9 to -4.3 mm Hg; p=0.004 - see figure 1), mean pressure (B= -9.1 mm Hg/100g; 95% CI: -15.4 to -2.8 mm Hg; p=0.009), but not systolic blood pressure. These associations remained after adjusting for gestation, between-pair difference in birthweight, neonatal morbidity and diet, current body mass index and immediate postnatal weight loss (partly due to physiological contraction of extra-cellular volume); and after expressing neonatal growth as change in postnatal weight z-score to 2 (or 3) weeks. Conversely, birthweight difference was unrelated to blood pressure difference at 8 or 13-16 years. Post-natal growth, corresponding to that in the third trimester, showed striking association with decreased later blood pressure independent of birthweight, gestation and key potential confounding factors. These novel data suggest neonatal factors programme blood pressure. Interestingly, only at 13-16 and not 8 years, was blood pressure related to neonatal growth, possibly because early influences on blood pressure are amplified after puberty3 or with increasing age4. Comparison between twins effectively controls for confounding by maternal nutrition and socio-economic factors1. Whether optimising neonatal growth in full term infants favourably programmes later blood pressure now merits investigation. Within-Pair Difference in Diastolic Blood Pressure of Adolescent Twins Against Within-pair Difference in Growth in the First 2 weeks After Birth Atul Singhal MD MRCP Alan Lucas MD FRCP From: The MRC Childhood Nutrition Research Centre, Institute of Child Health, 30 Guilford Street, London. Address for correspondence: Dr A Singhal, E mail: a.singhal@ich.ucl.ac.uk Funding: Medical Research Council References 1. Dwyer T, Blizzard L, Morely R, Ponsonby AL. Within pair association between birth weight and blood pressure at age 8 in twins from a cohort study. BMJ 1999; 319: 19: 1325-9. 2. Lucas A, Morley R. Does early nutrition in infants born before term programme later blood pressure. BMJ 1994; 309: 304-8. 3. Taittonen L, Nuutinen M, Turtinen J, Uhari M. Prenatal and postnatal factors in predicting later blood pressure among children: cardiovascular risk in young Finns. Pediatric Res 1996; 40: 627-32. 4. Law CM, Shiell AW. Is blood pressure inversely related to birth weight? The strength of evidence from a systematic review of the literature. J Hypertens 1996; 14: 935-41.
Why not report the result about diastolic and mean blood pressure? 19 April 2002
Clayman ZK Zhang, Resear staff MRC Environmental Epidemiology Unit,Southampton,SO16 6YD Send response to journal: Re: Why not report the result about diastolic and mean blood pressure?	Sir: Dr Dwyer's research is impresive to me,because it have perhaps been the first report ever tried to defend the 'fetal origin hypothesis' by way of sibling analysis.But I don't know why Dr Dwyer did not show any result of the correlation of birth size to diastolic blood pressure or even mean blood pressure.We certainly cannot imagin Dr Dwyer just measured the systolic blood pressure in this study.It perhaps because these correlations are simply did not statistically signicant.But it in fact also makes meaning,because systolic,diastolic and mean blood pressure reflect diffrent function of cardiovascular system,and the underlying mechinisms determining the relations of birth size and these blood indexex are also possibly different.In order to give a whole profile of the research,the reporter should not try to omit some fact sheets from the paper.