birth weight and subsequent blood pressure: a meta-analysis

15
Archives of Cardiovascular Disease (2012) 105, 99—113 Available online at www.sciencedirect.com REVIEW Birth weight and subsequent blood pressure: A meta-analysis Poids à la naissance et pression artérielle : méta-analyse Min Mu, Su-Fang Wang , Jie Sheng , Yan Zhao , Hu-Zhong Li, Chuan-Lai Hu, Fang-Biao Tao School of Public Health, Anhui Medical University, Meishan road 69, Hefei 230032, Anhui, China Received 22 August 2011; received in revised form 4 October 2011; accepted 4 October 2011 Available online 14 February 2012 KEYWORDS Birth weight; Hypertension; Blood pressure; Meta-analysis Summary Hypertension is becoming an important health problem in many countries. The ‘small baby syndrome hypothesis’ suggests that an inverse linear relationship exists between birth weight and later risk of hypertension; however, this relationship is under debate. We conducted a meta-analysis to examine the association between birth weight and subsequent blood pressure. Among 78 studies reporting on the association between birth weight and subse- quent blood pressure, 20 articles (reporting 27 original studies) were eligible for inclusion. Low birth weight (< 2500 g) compared with birth weight greater than 2500 g was associated with an increased risk of hypertension (odds ratio [OR] 1.21; 95% confidence interval [CI] 1.13, 1.30); high birth weight (> 4000 g) compared with birth weight less than 4000 g was associated with a decreased risk of hypertension (OR 0.78; 95% CI 0.71, 0.86). When low birth weight (< 2500 g) was compared with birth weight greater than 2500 g, mean systolic blood pressure (SBP) increased by 2.28 mmHg (95% CI 1.24, 3.33); when high birth weight (> 4000 g) was compared with birth weight less than 4000 g, mean SBP decreased by 2.08 mmHg (95% CI —2.98, —1.17). These find- ings indicate that there is an inverse linear association between birth weight and later risk of hypertension, and that this association primarily exists between birth weight and SBP. © 2011 Elsevier Masson SAS. All rights reserved. Abbreviations: BW, Birth weight; CI, Confidence interval; CVD, Cardiovascular disease; DBP, Diastolic blood pressure; OR, Odds ratio; SBP, Systolic blood pressure. Corresponding author. Fax: +86 551 3869179. E-mail address: [email protected] (S.-F. Wang). 1875-2136/$ see front matter © 2011 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.acvd.2011.10.006

Upload: min-mu

Post on 23-Nov-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Birth weight and subsequent blood pressure: A meta-analysis

Archives of Cardiovascular Disease (2012) 105, 99—113

Available online at

www.sciencedirect.com

REVIEW

Birth weight and subsequent blood pressure:A meta-analysis

Poids à la naissance et pression artérielle : méta-analyse

Min Mu, Su-Fang Wang ∗, Jie Sheng, Yan Zhao,Hu-Zhong Li, Chuan-Lai Hu, Fang-Biao Tao

School of Public Health, Anhui Medical University, Meishan road 69, Hefei 230032, Anhui,China

Received 22 August 2011; received in revised form 4 October 2011; accepted 4 October 2011Available online 14 February 2012

KEYWORDSBirth weight;Hypertension;Blood pressure;Meta-analysis

Summary Hypertension is becoming an important health problem in many countries. The‘small baby syndrome hypothesis’ suggests that an inverse linear relationship exists betweenbirth weight and later risk of hypertension; however, this relationship is under debate. Weconducted a meta-analysis to examine the association between birth weight and subsequentblood pressure. Among 78 studies reporting on the association between birth weight and subse-quent blood pressure, 20 articles (reporting 27 original studies) were eligible for inclusion. Lowbirth weight (< 2500 g) compared with birth weight greater than 2500 g was associated with anincreased risk of hypertension (odds ratio [OR] 1.21; 95% confidence interval [CI] 1.13, 1.30);high birth weight (> 4000 g) compared with birth weight less than 4000 g was associated with adecreased risk of hypertension (OR 0.78; 95% CI 0.71, 0.86). When low birth weight (< 2500 g) wascompared with birth weight greater than 2500 g, mean systolic blood pressure (SBP) increasedby 2.28 mmHg (95% CI 1.24, 3.33); when high birth weight (> 4000 g) was compared with birthweight less than 4000 g, mean SBP decreased by 2.08 mmHg (95% CI —2.98, —1.17). These find-

ings indicate that there is an inverse linear association between birth weight and later risk ofhypertension, and that this association primarily exists between birth weight and SBP.

All rights reserved.

© 2011 Elsevier Masson SAS.

Abbreviations: BW, Birth weight; CI, Confidence interval; CVD, Cardiovascular disease; DBP, Diastolic blood pressure; OR,Odds ratio; SBP, Systolic blood pressure.

∗ Corresponding author. Fax: +86 551 3869179.E-mail address: [email protected] (S.-F. Wang).

1875-2136/$ — see front matter © 2011 Elsevier Masson SAS. All rights reserved.doi:10.1016/j.acvd.2011.10.006

Page 2: Birth weight and subsequent blood pressure: A meta-analysis

100 M. Mu et al.

MOTS CLÉSHypertensionartérielle ;Pression artérielle ;Méta-analyse

Résumé. — L’hypertension artérielle est un problème de santé publique intéressant de nom-breux pays dans le monde. L’hypothèse du syndrome du petit bébé suggère une relation inverseentre le poids à la naissance et le risque ultérieur d’hypertension artérielle. Cependant, cettehypothèse reste controversée. Nous avons réalisé une méta-analyse examinant l’associationentre le poids à la naissance et la pression artérielle lors de la vie. Parmi 78 études rapportantune association entre poids à la naissance et pression artérielle, 20 articles rapportant 27 étudesoriginales ont été considérés dans cette méta-analyse. Un poids à la naissance inférieur à 2500 gcomparé à un poids à la naissance supérieur à 2500 g est associé à une augmentation de risqued’hypertension artérielle (odds ratio 1,21 ; intervalle de confiance [IC] 95 % 1,13—1,30) ; unpoids à la naissance supérieur à 4000 g comparé à un poids à la naissance inférieur à 4000 gest associé avec une diminution du risque d’hypertension artérielle (odds ratio 0,78 ; IC 95 %0,71—0,86). Lorsque le poids à la naissance est inférieur à 2500 g, comparativement à unpoids à la naissance supérieur à 2500 g, la pression artérielle systolique moyenne augmente de2,28 mmHg (IC 95 % 1,24—3,33) ; lorsque le poids à la naissance est supérieur à 4000 g, compar-ativement aux sujets dont le poids à la naissance est inférieur à 4000 g, la pression artériellesystolique moyenne diminue de 2,08 mmHg (IC 95 % —2,98, —1,17). Ces observations issuesde cette méta-analyse indiquent qu’il existe une association linéaire inverse entre le poids àla naissance et le risque ultérieur d’hypertension artérielle et cette association concerne enparticulier le poids à la naissance et la pression artérielle systolique.© 2011 Elsevier Masson SAS. Tous droits réservés.

I

HC[mptpseasttdaih(ah

dicaioaaci

B

whlaaabtetvab

M

L

Pat‘dsra

S

Ti

ntroduction

ypertension is one of the most important risk factors forVD [1] and the second leading cause of death in China2]. In China, people with hypertension were five timesore likely to develop stroke than those with normal bloodressure [3]. In Western populations, 30% of total mor-ality attributable to CVD could be prevented if bloodressure could be reduced by 10 mmHg [4]. Although exten-ively studied, the aetiology of hypertension cannot be fullyxplained by genetic factors and adulthood risk factors, suchs age, body mass index, physical activity and cigarettemoking [3,5]. It has been suggested that factors relatedo the intrauterine and postnatal environments may con-ribute to the development of hypertension [5]. Intrauterineevelopment may ‘programme’ foetal anatomy, physiologynd metabolism, and thus may affect the risk of diseases,ncluding hypertension, in later life [6—8]. This hypothesisas been evaluated in animal models and low birth weightBW) or intrauterine nutrition deficiency was found to bessociated with high blood pressure and/or resulted in adultypertension [9—11].

Barker et al. have suggested that there may beevelopmental periods in utero and in infancy during whichnappropriate nutrition and poor growth lead to long-termonsequences for adult CVD [12,13]. Blood pressure indults is typically inversely associated with BW [12,14]. Thisnverse relationship has also been described in childhood,nce adjustment is made for current size [15]. Subsequently,

number of studies found an association between low BWnd increased risk of adult hypertension. Most researcherslaimed that babies who were born short and thin had an

ncreased risk of hypertension in subsequent life [16—18].

Many authors have claimed that the relationship betweenW and later blood pressure is inversely linear [17,19—30],

Bcc

hich implies that high BW would lead to a decreased risk ofypertension. However, some researchers found that BW hadittle effect on subsequent blood pressure [32,33] or foundn increased risk of hypertension in subjects with low BWnd those with high BW [34,35]. Some studies that foundn inverse linear association between BW and subsequentlood pressure used a classification of BW that differed tohat in the present study. In one meta-analysis by Huxleyt al. [36], mean BW was used. We divided BW into lesshan 2500 g versus greater than 2500 g and greater than 4000ersus less than 4000 g, thus performing a meta-analysis with

different classification to further validate the relationshipetween BW and subsequent blood pressure.

ethods

iterature search strategy

ubMed and ScienceDirect databases were searched forrticles published between 1995 and 2011. We used theerms ‘birth weight’, ‘intrauterine growth restriction’,hypertension’, ‘blood pressure’, and ‘cardiovasculariseases’ in the full-text option. Furthermore, we manuallyearched all references cited in the original studies andeviews identified. In total, 78 studies reported on thessociation between BW and subsequent blood pressure.

tudies included and excluded criteria

o be eligible, a study had to fulfil the following criteria:t had to be an original report on the relationship between

W and blood pressure; and odds ratios [ORs] and 95%onfidence intervals [CIs] (or the data with which toalculate them) for hypertension or blood pressure values
Page 3: Birth weight and subsequent blood pressure: A meta-analysis

ag

S

SvmetoewscttwC

R

O

Birth weight and subsequent blood pressure

had to be presented. Alternatively, BWs could be reportedas categorical data with a certain range (e.g., < 2500 g,> 2500 g, > 4000 g, < 4000 g, etc.). Papers were excluded forthe following reasons:• if the title and abstract clearly did not contain data on

BW and hypertension or blood pressure;• if there were insufficient dichotomous data on BW and

hypertension or blood pressure;• duplicate;• if there was no measure of BW;• if the paper was a review or commentary article;• if the paper reported data using different BW categories.

Quality assessment

The Newcastle-Ottawa quality assessment scale was usedfor quality assessment [37]. Ten questions were assessedand each satisfactory answer received 1 point, resulting ina maximum score of 10. Only studies for which the majorityof questions were deemed satisfactory (i.e. with a score of6 or higher) were considered to be of high methodologicalquality.

Assessment of heterogeneity

The chi-square test was used to test for heterogeneity acrossstudies. A random-effects model was used to account forpossible heterogeneity between studies, which defaulted to

s

Aa

Figure 1. Flow chart of article screening and selection process. BP: bl

101

fixed-effects model approach in the absence of hetero-eneity [38]. A P value < 0.05 was considered significant.

tatistical analysis

tatistical analysis was conducted using Review Manager,ersion 5.0 (Nordic Cochrane Centre, Copenhagen, Den-ark). ORs were pooled for dichotomous outcomes from

ach study, Means ± standard errors were pooled for con-inuous variables from each study and the 95% CI for eachutcome was estimated to reflect the uncertainty of pointstimates. Sensitivity analysis was performed to determinehether differences in statistical methods, study design,

ample size and quality grade of the study affected studyonclusions. Publication bias was assessed by inspection ofhe funnel plot and by formal testing for funnel plot asymme-ry using Begg’s test and Egger’s test [31]. These calculationsere carried out using Stata/SE, version 10 (Stata Corp.,ollege Station, TX, USA).

esults

verview of studies included in the

ystematic review and meta-analysis

mong 78 studies reporting on the association between BWnd subsequent blood pressure, the following were excluded

ood pressure; BW: birth weight.

Page 4: Birth weight and subsequent blood pressure: A meta-analysis

102

M.

Mu

et al.

Table 1 Characteristics of 20 studies included in the meta-analysis (1998—2011).

Study Country Studydesign

Year of studybaseline

Total numberof subjects

Age(years)

Race BW ascertainmentmethod

BW reference categoryfor adjusted estimate

Outcome

Tamakoshi et al.2006 [35]

Japan Cohort 2002 2303 men, 804women

33—66 Yellow Questionnaire < 2500 g;2500—3500 g; > 3500 g

Hypertension;BP

Lurbe et al.2001 [25]

Spain Cohort 369 women 4—18 White Recorded frommedical records

< 2500 g; 2500—2999 g;3000—3299 g;3300—3599 g

BP

Eriksson et al.2000 [34]

Finland Cohort 1924—1933 975 men, 983women

6—16 White Recorded frommedical records

< 2500 g; 2500—3000 g;3001—3500 g;3501—4000 g; > 4000 g

Hypertension

Hardy et al.2003 [40]

England Cohort 1946 2815 men,2547 women

36—53 White Interview < 2500 g; 2500—3000 g;3001—3500 g;3501—4000 g; > 4000 g

BP

Fallkner et al.2004 [20]

USA Cohort 1988 133 men, 117women

11—14 White,Black,His-panic,Asian

Recorded frommedical records

< 2500 g; > 2500 g BP

Bergvall et al.2007 [17]

Sweden Cohort 1926—1958 15 White Questionnaire < 1999 g; 2000—2499 g;2500—2999 g;3000—3499 g; ≥ 3500 g

Hypertension

Barker et al.2002 [39]

Finland Cohort 1933—1944 4627 men,4130 women

12 White Recorded frommedical records

< 3000 g; 3000—3500 g;3500—4000 g; > 4000 g

Hypertension

Yliharsila et al.2003 [30]

Finland Cohort 1924—1933 7707 men,8558 women

70 White Recorded frommedical records

< 2500 g; 2500—3000 g;3000—3500 g;3500—4000 g; > 4000 g

Hypertension

Yarbrough et al.1998 [29]

USA Cohort 1984 and 1987 303 women 50—84 White Questionnaire Three predefined BWcategories in pounds

BP

Kistner et al.2000 [23]

Sweden Case-Control

1970 and 1974 50 women 18 White Recorded frommedical records

Percentile for gestation BP

Wei et al. 2003[33]

China Case-control

1992—1997 198 men, 231women

6—18 Yellow Telephone,questionnaire

< 2500 g; 2500—2999 g;3000—3499 g;3500—3999 g; > 4000 g

BP

Tian et al. 2006[28]

China Cross-sectional

1992 373 men, 600women

46.2 Yellow Questionnaire < 2500 g; 2500—2999 g;3000—3499 g;3500—3999 g; > 3500 g

Hypertension;BP

Pyhala et al.2009 [26]

Finland Case-control

35 men, 46women

23.1 White Recorded frommedical records

< 2500 g; > 2500 g BP

Page 5: Birth weight and subsequent blood pressure: A meta-analysis

Birth w

eight and

subsequent blood

pressure

103

Table 1 (Continued )

Study Country Studydesign

Year of studybaseline

Total numberof subjects

Age(years)

Race BW ascertainmentmethod

BW reference categoryfor adjusted estimate

Outcome

Jarvelin et al.2004 [22]

Finland Cohort 1966 2858 men,3102 women

31 White Questionnaire < 2500 g; 2500—2999 g;3000—3499 g;3500—3999 g;4000—4499 g; ≥ 4500 g

BP

Liew et al. 2008[24]

USA Cohort 1987—1989 5—72 White Interview < 2500 g; 2500—2999 g;3000—3499 g;3500—3999 g;4000—4499 g; ≥ 4500 g

BP

Koupil et al.2005 [41]

Sweden Cohort 1970 736 men 70 White Recorded frommedical records

1400—3290 g;3300—3600 g;3610—3960 g;3970—5400 g

BP

Liew et al. 2008[32]

USA Cohort 1987—1989 4286 men,5444 men

45—64 Black,White

Interview < 2500 g; 2500 — 4000 g;> 4000 g

Hypertension;BP

Salgado et al.2009 [27]

Brazil Case-control

33 men, 35women

8—11 Black Recorded fromchild’s health card

≤ 2500 g; ≥ 2500 g BP

Hirschler et al.2008 [21]

USA Cross-sectional

2006—2007 511 men, 516women

10 White Recorded frommedical records

< 2500 g; 2500—4000 g;> 4000 g

SBP

Zhao et al. 2002[42]

China Cohort 1997—2000 75049 women 40—70 Yellow Interview < 2500 g; 2500—3249 g;3250—3999 g; > 4000 g

Hypertension

BP: blood pressure; BW: birth weight; SBP: systolic blood pressure

Page 6: Birth weight and subsequent blood pressure: A meta-analysis

104 M. Mu et al.

Figure 2. Odds ratios for hypertension in subjects with low birth weight (< 2500 g) compared with subjects with birth weight greater than2 rst auo n par

fdpddmc2(dsc[

taiiow[

M

Nwgut

b1

jBeohP

(sitsm—

batP(B—

500 g, in a meta-analysis. Studies are ordered alphabetically by fif a fixed-effects model; 95% confidence intervals (CIs) are shown i

or the following reasons (Fig. 1): title and abstract clearlyid not contain data on BW and hypertension or bloodressure (n = 24); did not provide sufficient dichotomousata on BW and hypertension or blood pressure (n = 12);uplicates (n = 5); no measure of BW (n = 7); reviews or com-entary articles (n = 7); reported data using different BW

ategories (n = 3). A total of 20 original articles reporting7 original studies were identified [17,20—30,32—35,39—42]one article [40] described six studies and one article [17]escribed three studies); the studies comprised 14 cohorttudies [17,20,22,24,25,29,30,32,34,35,39—42], four case-ontrol studies [23,26,27,33] and two cross-sectional studies21,28]. Study characteristics are displayed in Table 1.

The quality of each study, in terms of popula-ion and sampling methods, description of exposurend outcomes, and statistical adjustment of the data,s summarized in Appendix 1. Thirteen articles report-ng 20 original studies received scores of 6 or highern the Newcastle-Ottawa quality assessment scale andere considered to be of high methodological quality

17,20,22,24,25,29,30,32,33,35,40—42] (Appendix 1).

eta-analysis

ine studies analysed the risk of hypertension in subjects

ith low BW (< 2500 g) compared with subjects with BWreater than 2500 g. Data from these studies were assessedsing the fixed-effects model (�2 = 5.20; P = 0.74; I2 = 0%) andhe results of this analysis revealed a positive association

S

Tw

thor. The pooled odds ratios (diamond) were calculated by meansentheses and as horizontal bars. M-H: Mantel-Haenszel.

etween low BW and hypertension (OR 1.21; 95% CI 1.13,.30; P < 0.0001; Fig. 2).

Fig. 3 shows the forest plot for risk of hypertension in sub-ects with high BW (> 4000 g) compared with subjects withW less than 4000 g. This effect was assessed using the fixed-ffects model (�2 = 2.43; P = 0.66; I2 = 0%) and the resultsf this analysis revealed a negative association betweenigh BW and hypertension (OR 0.78; 95% CI 0.71, 0.86;

< 0.0001).Figs. 4 and 5 indicated that subjects with low BW

< 2500 g) had a higher systolic blood pressure (SBP) thanubjects with BW greater than 2500 g (the total mean SBPncreased by 2.28 mmHg; 95%CI 1.24, 3.33; P < 0.0001) andhat subjects with high BW (> 4000 g) had a lower SBP thanubjects with BW less than 4000 g in the random-effectsodel (the total mean SBP decreased by 2.08 mmHg; 95%CI2.98, —1.17; P < 0.0001).

There was a weaker association between BW and diastoliclood pressure (DBP) (Figs. 6 and 7): low BW (< 2500 g) wasssociated with increased DBP compared with BW greaterhan 2500 g (mean increase 1.01 mmHg; 95% CI 0.19, 1.83;

= 0.02) in the random-effects model. In contrast, high BW> 4000 g) was associated with decreased DBP compared withW less than 4000 g (mean decrease —0.37 mmHg; 95% CI1.19, 0.45; P = 0.37) in the random-effects model.

ensitivity analysis

o further investigate the association between BW and SBP,e performed a sensitivity analysis. However, age, the

Page 7: Birth weight and subsequent blood pressure: A meta-analysis

Birth weight and subsequent blood pressure 105

Figure 3. Odds ratios for hypertension in subjects with high birth weight (> 4000 g) compared with subjects with birth weight less than4000 g, in a meta-analysis. Studies are ordered alphabetically by first author. The pooled odds ratios (diamond) were calculated by meansof a fixed-effects model; 95% confidence intervals (CIs) are shown in parentheses and as horizontal bars. M-H: Mantel-Haenszel.

Figure 4. The mean differences in systolic blood pressure in subjects with low birth weight (< 2500 g) compared with subjects withbirth weight greater than 2500 g, in a meta-analysis. Studies are ordered alphabetically by first author. The pooled mean differences were

calculated by means of a random-effects model; 95% confidence intervalvariance; SD: standard deviation.

s (CIs) are shown in parentheses and as horizontal bars. IV: inverse

Page 8: Birth weight and subsequent blood pressure: A meta-analysis

106 M. Mu et al.

Figure 5. The mean differences in systolic blood pressure in subjects with high birth weight (> 4000 g) compared with subjects with birthweight less than 4000 g, in a meta-analysis. Studies are ordered alphabetically by first author. The pooled mean differences were calculatedby means of a random-effects model; 95% confidence intervals (CIs) are shown in parentheses and as horizontal bars. IV: inverse variance;SD: standard deviation.

Figure 6. The mean differences in diastolic blood pressure in subjects with low birth weight (< 2500 g) compared with subjects withbirth weight greater than 2500 g, in a meta-analysis. Studies are ordered alphabetically by first author. The pooled mean differences werecalculated by means of a random-effects model; 95% confidence intervals (CIs) are shown in parentheses and as horizontal bars. IV.: inversevariance; SD: standard deviation.

Page 9: Birth weight and subsequent blood pressure: A meta-analysis

Birth weight and subsequent blood pressure 107

Figure 7. The mean differences in diastolic blood pressure in subjects with high birth weight (> 4000 g) compared with subjects with birthweight less than 4000 g, in a meta-analysis. Studies are ordered alphabetically by first author. The pooled mean differences were calculated

are

sivh

hhmtc1iprctsatmlBtihmbtB

by means of a random-effects model; 95% confidence intervals (CIs)SD: standard deviation.

method of obtaining data on BW, race and study designhad an effect on the BW/SBP association. When comparinglow BW (< 2500 g) with BW greater than 2500 g, the BW/SBPassociation was more obvious when BW was recorded frommedical records, when age was greater than 35 years, incase-control studies and in studies of white subjects. In con-trast, when comparing high BW (> 4000 g) with BW less than4000 g, the BW/SBP association was more obvious when BWwas recorded from an interview, when age was greater than35 years, in cohort studies and in cross-sectional studies. Asthese variables had a strong impact on the BW/SBP associ-ation, their differences may partially explain the observedheterogeneity between studies (Appendix 2).

Publication bias

Inspection did not reveal an obvious effect of publicationbias and Egger’s test for publication bias was not statisticallysignificant (P = 0.175 for studies comparing BW < 2500 g withBW < 2500 g, Appendix 3; P = 0.452 for studies comparingBW > 4000 g with BW < 4000 g, Appendix 4).

Discussion

Many studies have reported an association between BW andlater diseases, including diabetes, coronary heart diseaseand hypertension. Some scholars presume that there is aU-shaped association between BW and subsequent risk of

type 2 diabetes [33,43,44], and that there is an inverselinear association between BW and later risk of hyperten-sion [35,42], That is to say, as BW increases, the prevalenceof adult hypertension decreases gradually. In our study,

S

dl

shown in parentheses and as horizontal bars. IV.: inverse variance;

ubjects with high BW had lower hypertension prevalencen later life but more studies are needed to confirm theiew that high BW is associated with decreased risk of laterypertension.

The reason why low BW would increase the risk of adultypertension is not yet clear. Although some investigationsave not detected evidence to support the low BW theory,ost reports support this concept to some degree. Reviews

hat have examined the results of multiple studies that,ollectively, included thousands of subjects, estimate that a

kg increase in BW is associated with a 2—4 mmHg reductionn SBP [14,36]. Based on their reviews, these authors haveroposed that the findings of statistically significant inverseelationships between BW and later blood pressure may beonfounded by the effects of random error, particularly inhe ascertainment of BW. Other confounders included theelective emphasis of particular results and inappropriatedjustment for current weight. These authors concludedhat BW, as a surrogate measure of the intrauterine environ-ent, had little, if any, relationship to blood pressure level

ater in life. Low BW is a marker of poor foetal nutrition.arker et al. suggested that the foetus would reprogramo adapt to deleterious situations in order to survive poorntrauterine nutrition, which would increase the risk ofaving hypertension in later life [45,46]. Another suggestedechanism is that common genetic factors could underlieoth foetal development and later disease risk or relatedraits, thereby producing an inverse association betweenW and SBP; this hypothesis was supported by a recent

wedish study [47].

Epidemiological evidence accumulated over the past twoecades has demonstrated a significant association betweenow BW and subsequent adult hypertension, diabetes and

Page 10: Birth weight and subsequent blood pressure: A meta-analysis

1

Crhed

pNltoahbFw

evawida

ot

C

Ica

D

Tc

A

Tao

S

C

C

C

08

VD [19,48,49]. Suboptimal intrauterine nutrition wouldestrict foetal growth and result in lower BW. Thus, low BWas become the clinical marker of an adverse intrauterinenvironment and a possible risk factor for future chroniciseases [50,51].

The principal limitation of this study was the use ofotentially biased evidence. No studies met all of theewcastle-Ottawa criteria. However, 20/27 studies met at

east 6/10 criteria. Sensitivity analyses also demonstratedhat the study quality (scored as total number of pointsut of 10) had a strong impact on the BW/blood pressuressociation. Low quality grade studies increased interstudyeterogeneity. Some studies reported a correlation for menut not women, which was inconsistent with our findings.urther research is needed to unequivocally determinehether sex affects the BW/blood pressure association.

In the included studies, BW was determined using sev-ral different methods, ranging from measured or reportedalues at birth or shortly after birth to questionnairesnd interviews administered at different postnatal agesith different recall periods. As a consequence, the data

ncluded in this meta-analysis may suffer from differingegrees of completeness and accuracy. We have performed

sensitivity analysis to compensate for the heterogeneity

r

A

tudies Selection

1 2 3 4

ohortTamakoshi K, et al. 2006 * * *

Lurbe E, et al. 2001 * * *

Eriksson J, et al. 2000 * * *

Hardy R (1), et al. 2003 * * *

Hardy R (2), et al. 2003 * * *

Hardy R (3), et al. 2003 * * *

Hardy R (4), et al. 2003 * * *

Hardy R (5), et al. 2003 * * *

Hardy R (6), et al. 2003 * * *

Fallkner B, et al. 2004 * * * *

Bergvall N (1), et al. 2007 * * *

Bergvall N (2), et al. 2007 * * *

Bergvall N (3), et al. 2007 * * *

Barker D.J.P., et al. 2002 * * *

Yliharsila H, et al. 2003 * * *

Jarvelin M.R., et al. 2004 * * *

Liew G (1), et al. 2008 * * * *

Liew G (2), et al. 2008 * * * *

Koupil L, et al. 2005 * * * *

Zhao M.F., et al.2002 * * * *

ase-controlYarbrough D.E., et al. 1998 * * *

Kistner A, et al. 2000 *

Wei J.N., et al. 2003 * * * *

Pyhala R, et al. 2009 * * *

Salgado C.M., et al. 2009 *

ross-sectionalHirschler V, et al. 2008 * * *

Tian JY, et al. 2006 * *

M. Mu et al.

f the data; however, these limitations must be noted, andhe results should be considered with caution.

onclusions

n conclusion, we find that there is an inversely linear asso-iation between BW and risk of adult hypertension. Thisssociation primarily appears to be between BW and SBP.

isclosure of interest

he authors declare that they have no conflicts of interestoncerning this article.

cknowledgements

his work was supported by the Danone Institute, China. Were very grateful to the Department of Nutrition, the Schoolf Public Health, and Anhui Medical University, for encou-agement and helpful suggestions relating to this study.

ppendix 1. Assessment of study quality.

Comparability Outcome

5A 5B 6 7 8 Score

* * ** * * ********** * * * *******

* * ****** * * * ******** * * ******** * * ******** * * ******** * * ******** * * * ******** * * * *******

* * * ******* * ** * ********* * ** * ********

* * ****** * ** * * ********** * * ******* * * * ********* * * * ********

* * * ******* * * * ********

* ** * ******** * * * ****** * * * ********

* * ****** * ***

* * ****** * ****

Page 11: Birth weight and subsequent blood pressure: A meta-analysis

Birth weight and subsequent blood pressure 109

Appendix 2. Birth weight and systolic blood pressure (SBP): sensitivity analysis (randomeffects model).

Study characteristics > 2500 g and < 2500 g > 4000 g and < 4000 gMean ± SE (95% CI) Mean ± SE (95% CI)

Age> 35 2.18 (1.01, 3.34) —2.26 (—3.25, —1.27)< 35 1.99 (—1.03, 5.02) —1.27 (—3.20, 0.65)

Source of data for birth weightQuestionnaire 1.14 (—1.19, 3.47) 0.07 (—3.03, 3.18)Interview 1.73 (0.05, 3.41) —2.51 (—3.15, —1,50)Recorded from medical records 3.12 (—0.82, 7.05) 0.32 (—4.38, —3.03)

RaceWhite 2.51 (0.96, 4.06) —2.18 (—3.15, —1.21)Yellow and other 1.58 (—0.35, 3.50) —0.73 (—5.16, 3.70)

Study designCohort 2.22 (1.10, 33) —2.23 (—3.17, —1.29)Case-control 6.01 (2.31, 70) 2.00 (—2.03, —6.03)Cross-sectional —3.90 (—6.34, —1.47) 2.01 (—5.28, —1.08)

Appendix 3. The Egger’s test BW < 2500 g vs. BW > 2500 g.

OR SBP DPB

Egger’s test 0.175 0.255 0.556

Page 12: Birth weight and subsequent blood pressure: A meta-analysis

1

10 M. Mu et al.
Page 13: Birth weight and subsequent blood pressure: A meta-analysis

Birth weight and subsequent blood pressure 111

Appendix 4. The Egger’s test BW > 4000 g vs. BW < 4000 g.

OR SBP DPB

Egger’s test 0.452 0.543 0.665

Page 14: Birth weight and subsequent blood pressure: A meta-analysis

1

R

[

12 M. Mu et al.

eferences

[1] Wright JT, Hammonds VC. Hypertension: Epidemiology andcontemporary management strategies. In: Wong ND, BlackHR, Gardin JM, editors. Preventive cardiology. New-York, NY:McGraw-Hill Professional; 1999. p. 133—64.

[2] Chen XS, Ge KY. Nutrition transition in China: the growth ofaffluent diseases with the alleviation of undernutrition. AsiaPacific J Clin Nutr 1995;4:287—93.

[3] Ueshima H, Zhang XH, Choudhury SR. Epidemiology of hyper-tension in China and Japan. J Hum Hypertens 2000;14:765—9.

[4] Rose G. Sick individuals and sick populations. Int J Epidemiol1985;14:32—8.

[5] Barker DJ. Rise and fall of Western diseases. Nature1989;338:371—2.

[6] Barker DJ. Maternal nutrition, fetal nutrition, and disease inlater life. Nutrition 1997;13:807—13.

[7] Byrne CD, Phillips DI. Fetal origins of adult disease: epidemiol-ogy and mechanisms. J Clin Pathol 2000;53:822—8.

[8] Godfrey KM, Barker DJ. Fetal nutrition and adult disease. AmJ Clin Nutr 2000;71:1344S—52S.

[9] Langley-Evans SC, Phillips GJ, Gardner DS, et al. Role of

[11] Persson E, Jansson T. Low birth weight is associated with ele-vated adult blood pressure in the chronically catheterizedguinea-pig. Acta Physiol Scand 1992;145:195—6.

[12] Barker DJ, Osmond C, Golding J, et al. Growth in utero, bloodpressure in childhood and adult life, and mortality from car-diovascular disease. BMJ 1989;298:564—7.

[13] Barker DJ, Winter PD, Osmond C, et al. Weight in infancyand death from ischaemic heart disease. Lancet 1989;2:577—80.

[14] Law CM, Shiell AW. Is blood pressure inversely related to birthweight? The strength of evidence from a systematic review ofthe literature. J Hypertens 1996;14:935—41.

[15] Whincup PH. Fetal origins of cardiovascular risk: evidence fromstudies in children. Proc Nutr Soc 1998;57:123—37.

[16] Barker DJ, Osmond C. Infant mortality, childhood nutrition,and ischaemic heart disease in England and Wales. Lancet1986;1:1077—81.

[17] Bergvall N, Iliadou A, Johansson S, et al. Genetic andshared environmental factors do not confound the associationbetween birth weight and hypertension: a study among Swedishtwins. Circulation 2007;115:2931—8.

[18] Martyn CN, Barker DJ, Jespersen S, et al. Growth in utero,

glucocorticoids in programming of maternal diet-inducedhypertension in the rat. J Nutr Biochem 1996;7:173—8.

10] Manning J, Vehaskari VM. Low birth weight-associated adulthypertension in the rat. Pediatr Nephrol 2001;16:417—22.

[

adult blood pressure, and arterial compliance. Br Heart J

1995;73:116—21.

19] Barker D. Commentary: birthweight and coronary heart diseasein a historical cohort. Int J Epidemiol 2006;35:886—7.

Page 15: Birth weight and subsequent blood pressure: A meta-analysis

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[51] Franco MC, Christofalo DM, Sawaya AL, et al. Effects of

Birth weight and subsequent blood pressure

[20] Falkner B, Hulman S, Kushner H. Effect of birth weight onblood pressure and body size in early adolescence. Hyperten-sion 2004;43:203—37.

[21] Hirschler V, Bugna J, Roque M, et al. Does low birth weightpredict obesity/overweight and metabolic syndrome in ele-mentary school children? Arch Med Res 2008;39:796—802.

[22] Jarvelin MR, Sovio U, King V, et al. Early life factors and bloodpressure at age 31 years in the 1966 northern Finland birthcohort. Hypertension 2004;44:838—46.

[23] Kistner A, Celsi G, Vanpee M, et al. Increased blood pressure butnormal renal function in adult women born preterm. PediatrNephrol 2000;15:215—20.

[24] Liew G, Wang JJ, Klein R, et al. The relationshipbetween birthweight and early age-related maculopathy: theatherosclerosis risk in communities study. Ophthalmic Epi-demiol 2008;15:56—61.

[25] Lurbe E, Torro I, Rodriguez C, et al. Birth weight influencesblood pressure values and variability in children and adoles-cents. Hypertension 2001;38:389—93.

[26] Pyhala R, Raikkonen K, Feldt K, et al. Blood pressure responsesto psychosocial stress in young adults with very low birthweight: Helsinki study of very low birth weight adults. Pedi-atrics 2009;123:731—4.

[27] Salgado CM, Jardim PC, Teles FB, et al. Low birth weight as amarker of changes in ambulatory blood pressure monitoring.Arq Bras Cardiol 2009;92:107—21.

[28] Tian JY, Cheng Q, Song XM, et al. Birth weight and risk of type 2diabetes, abdominal obesity and hypertension among Chineseadults. Eur J Endocrinol 2006;155:601—7.

[29] Yarbrough DE, Barrett-Connor E, Kritz-Silverstein D, et al. Birthweight, adult weight, and girth as predictors of the metabolicsyndrome in postmenopausal women: the Rancho Bernardostudy. Diabetes Care 1998;21:1652—8.

[30] Yliharsila H, Eriksson JG, Forsen T, et al. Self-perpetuatingeffects of birth size on blood pressure levels in elderly people.Hypertension 2003;41:446—50.

[31] Begg CB, Mazumdar M. Operating characteristics of arank correlation test for publication bias. Biometrics1994;50:1088—101.

[32] Liew G, Wang JJ, Duncan BB, et al. Low birthweight isassociated with narrower arterioles in adults. Hypertension2008;51:933—8.

[33] Wei JN, Sung FC, Li CY, et al. Low birth weight and highbirth weight infants are both at an increased risk to havetype 2 diabetes among schoolchildren in Taiwan. Diabetes Care2003;26:343—8.

[34] Eriksson J, Forsen T, Tuomilehto J, et al. Fetal and child-hood growth and hypertension in adult life. Hypertension

2000;36:790—4.

[35] Tamakoshi K, Yatsuya H, Wada K, et al. Birth weight and adulthypertension: cross-sectional study in a Japanese workplacepopulation. Circ J 2006;70:262—7.

113

36] Huxley R, Neil A, Collins R. Unravelling the fetal originshypothesis: is there really an inverse association betweenbirthweight and subsequent blood pressure? Lancet 2002;360:659—65.

37] Wells GA, Shea B, O’Connell D, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomizedstudies in meta-analyses. Accessed June 2009. http://www.ohri.ca/programs/clinical epidemiology/oxford.asp.

38] Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsis-tency in meta-analyses. BMJ 2003;327:557—60.

39] Barker DJ, Forsen T, Eriksson JG, et al. Growth and living condi-tions in childhood and hypertension in adult life: a longitudinalstudy. J Hypertens 2002;20:1951—6.

40] Hardy R, Kuh D, Langenberg C, et al. Birthweight, childhoodsocial class, and change in adult blood pressure in the 1946British birth cohort. Lancet 2003;362:1178—83.

41] Koupil I, Leon DA, Byberg L. Birth weight, hypertensionand ‘‘white coat’’ hypertension: size at birth in relation tooffice and 24-h ambulatory blood pressure. J Hum Hypertens2005;19:635—42.

42] Zhao M, Shu XO, Jin F, et al. Birthweight, childhood growthand hypertension in adulthood. Int J Epidemiol 2002;31:1043—51.

43] Dyck RF, Klomp H, Tan L. From ‘‘thrifty genotype’’ to ‘‘heftyfetal phenotype’’: the relationship between high birthweightand diabetes in Saskatchewan registered Indians. Can J PublicHealth 2001;92:340—4.

44] McCance DR, Pettitt DJ, Hanson RL, et al. Birth weight andnon-insulin dependent diabetes: thrifty genotype, thrifty phe-notype, or surviving small baby genotype? BMJ 1994;308:942—5.

45] Barker DJ. Fetal and infant origins of adult disease. Hoboken,New Jersey: Wiley-Blackwell; 1992.

46] Langley-Evans SC. Fetal programming of cardiovascular func-tion through exposure to maternal undernutrition. Proc NutrSoc 2001;60:505—13.

47] Barker DJ, Osmond C, Forsen TJ, et al. Trajectories of growthamong children who have coronary events as adults. N Engl JMed 2005;353:1802—9.

48] Barker DJ, Gluckman PD, Godfrey KM, et al. Fetal nutritionand cardiovascular disease in adult life. Lancet 1993;341:938—41.

49] Smith GD, Sterne J, Tynelius P, et al. Birth weight of offspringand subsequent cardiovascular mortality of the parents. Epi-demiology 2005;16:563—9.

50] Barker D. Mothers, babies and disease in later life. 2nd ed.London: BMJ Publishing Group; 1998.

low birth weight in 8 to 13-year-old children: implicationsin endothelial function and uric acid levels. Hypertension2006;48:45—50.