Introduction

Birth weight is an important indicator of the nutrition and growth status of fetuses. Poor nutrition in pregnancy often leads to low birth weight, whereas overnutrition leads to high birth weight (HBW).1, 2 Barker et al.3 showed that reduced fetal growth increases cardiovascular disease mortality in adults. The ‘fetal origin hypothesis’ of adult diseases was introduced to explain the relationship between low birth weight and adult diseases.4 Numerous studies on low birth weight have shown that birth weight is inversely related to the risk of metabolic syndrome,5, 6, 7, 8 which is defined as a combination of at least three of the following five components: obesity, high blood pressure, high serum triglycerides, low serum high-density lipoprotein and impaired glucose or insulin resistance.9 The ‘fetal origin hypothesis’ also suggests that an adverse intrauterine environment (for example, characterized by poor nutrition) may reduce fetal growth by programming metabolic development and thus lead to lifelong physiological changes that predispose the body to metabolic diseases.10

HBW, which is also termed macrosomia or large for gestational age in obstetrics, is defined as birth weight >4000 g or >90th percentile of gestational age.11 HBW is usually associated with maternal obesity, excessive gestational weight gain, or gestational diabetes mellitus.12 As the prevalence of maternal obesity and gestational diabetes mellitus has increased worldwide, the prevalence of HBW has also increased.13, 14

According to the ‘fetal origin hypothesis’, HBW may elicit effects on blood pressure that differ from those for individuals with low birth weight and normal birth weight (NBW; defined as birth weight between 2500 g and 4000 g); that is, subjects with HBW may have a lower risk of metabolic diseases than those with NBW. However, overnutrition and overweight in multiple stages of an individual’s life are key risk factors for metabolic syndromes. Therefore, conflicting hypotheses have been proposed to explain the relationship between HBW and blood pressure. However, most published studies have focused primarily on low birth weight without considering subjects with HBW, or they have mixed subjects with HBW into the NBW group, making the comparison more heterogeneous.7, 15, 16, 17, 18

We performed this systematic review to clarify the relationship between HBW and blood pressure, to explain the paradox of the ‘fetal origin hypothesis’ for HBW, and to assess the effects of HBW on blood pressure and hypertension in later life compared with individuals of NBW.

Methods

The systematic review and meta-analysis were conducted by following the Meta-Analysis of Observational Studies in Epidemiology Group checklist,19 a checklist that includes the processes by which the data are obtained, the analysis is performed, and the findings are reported in meta-analyses.

We performed a literature search in the PubMed and Embase databases using the following terms: (‘birth weight’ OR ‘birth size’ OR ‘macrosomia’ OR ‘large for gestational age’) NOT (‘preterm birth’ OR ‘low birth weight’ OR ‘small for gestational age’ OR ‘prematurity’ OR ‘twin’) AND (‘metabolic syndrome’ OR ‘hypertension’ OR ‘cardiovascular disease’ OR ‘blood pressure’). The search was limited to studies with human subjects that were published in the English language. The publication dates were before October 2012. Two authors screened the search results independently. The screened results from each reviewer were combined after deliberation. The full text articles were retrieved and checked for eligibility.

To be eligible for inclusion, studies had to satisfy the following criteria: (1) the study was an original report on the relationship(s) between birth weight and blood pressure or hypertension; (2) the subjects were singletons born at full-term; (3) the outcomes were investigated at ages >1 year; (4) HBW and NBW were distinctly defined with HBW as ⩾4000 g or ⩾the 90th percentile for gestational age and NBW as 2500–4000 g or the 10–90th percentiles for gestational age.

Furthermore, a manual search was conducted among the reference lists for all of the eligible studies and related reviews. The overall process is illustrated in Figure 1.

Figure 1
figure 1

Flow chart of literature review.

Full size image

From the studies that were considered eligible for this review, two authors collected information independently using standardized sheets. The means and the s.d. values for blood pressure were determined directly or by performing calculations. The numbers of participants with hypertension in each group were obtained for the relative risk (RR) calculations. If possible, all of the original data were extracted using the minimum subgroups.

The Newcastle-Ottawa quality assessment scale was used for quality assessment.20 Eight factors were assessed, and each satisfactory factor received one star (a maximum of nine stars were used).

We used STATA version 11.0 (Stata Corp., College Station, TX, USA) for the data analysis. The heterogeneity among the studies was assessed using Q and I2.21 The data were synthesized using a random-effects model or a fixed-effect model according to the heterogeneity. Meta-regression was used to confirm the heterogeneity and the relationship between the characteristics of the studies and the effects of HBW.

Results

Overall

We obtained 8756 records of studies from the PubMed and Embase databases in the computer search. After the abstracts of the studies were screened, 182 potentially relevant studies remained. We further assessed the details, performed a manual search, and determined that 31 studies satisfied the inclusion criteria. We primarily excluded studies because they lacked an HBW group. We also excluded those with inappropriate birth weight groups (that is, the cutoff points for HBW and NBW did not satisfy the inclusion criteria).

Blood pressure and/or hypertension data were obtained from 31 studies, and some of them were included in this review. In particular, six studies provided mean values but no standard deviations; another study provided Z scores. These seven studies were not included in the meta-analysis.

Four studies focused on overweight or obesity. Furthermore, at least one study included the following types of participants: offspring with gestational diabetes mellitus, children with abnormal urinalysis, individuals with type 1 diabetes and women with pregnancy-induced hypertension. The characteristics of the 31 studies are summarized in Table 1.

Table 1 Brief descriptions of the studies including in this review
Full size table

The quality of the studies in terms of subject selection, as well as the comparability and reliability of outcomes, is summarized in Table 2.

Table 2 Assessment of quality of the study
Full size table

Systolic blood pressure

Fourteen studies provided 24 mean differences (MDs) in SBP between individuals with HBW and NBW. The meta-analysis of random effects revealed an overall MD of –0.25 mm Hg (95% confidence interval, –0.92, 0.42) with considerable heterogeneity (I2=79.3%, Q-statistic: P<0.0001). The subgroup analysis by age (using the mean age from each study) showed that only the age group >40 years exhibited low heterogeneity. As the individuals grew older, the MD in SBP changed from positive to negative. The results indicate that HBW is associated with higher SBP in younger subjects, but lower SBP in older subjects (Figure 2).

Figure 2
figure 2

The differences of systolic blood pressure (SBP) between HBW and NBW in meta-analysis. Studies ordered by mean ages at which the outcome was measured. The pooled MDs were calculated by means of a random effects model; 95% confidence intervals (CI) are shown in parentheses and as horizontal bars. HBW, high birth weight, NBW, normal birth weight, WMD, weighted mean difference.

Full size image

To explore the sources of heterogeneity in the pooled MD, we performed a meta-regression analysis, which revealed that age explains the majority of the variance between the studies (value of adjusted R2). The regression coefficient also indicates that age was inversely associated with the MD in SBP (Table 3).

Table 3 Meta-regression for the MD of SBP between HBW and NBW
Full size table

Among the seven studies excluded from the meta-analysis, all five studies on adults showed that SBP was lower among individuals with HBW than individuals with NBW (statistical test unavailable).45, 46, 47, 48, 49 One of the two studies on children showed that SBP was higher among those with HBW than those with NBW (non-significant),23 in agreement with the results shown in Figure 2.

Diastolic blood pressure

Thirteen studies provided 23 MDs in DBP between those with HBW and those with NBW. The pooled MD was 0.20 mm Hg (95% confidence interval, –0.23, 0.62). The heterogeneity among these studies was also high (I2=72.6%, Q-statistic: P<0.0001). A subgroup analysis by age showed that the heterogeneity in each group was low. With increasing age, the MD in DBP changed from positive to negative. The results indicate that HBW is associated with higher DBP in younger subjects but lower DBP in older subjects (Figure 3).

Figure 3
figure 3

The differences of diastolic blood pressure (DBP) between HBW and NBW in meta-analysis. Studies ordered by the mean age at which the outcome was measured. The pooled MDs were calculated by means of a random-effects model; 95% confidence intervals (CI) are shown in parentheses and as horizontal bars. HBW, high birth weight; NBW, normal birth weight, WMD, weighted MD.

Full size image

The same characteristics of the subjects and the studies mentioned above were included in the meta-regression. The results of the meta-regression showed that age might be the main factor contributing to the significant heterogeneity. The regression coefficient also showed that age was inversely associated with the MD in DBP (Table 4).

Table 4 Meta-regression for the MD of DBP between HBW and NBW
Full size table

Of the two studies that were excluded from the meta-analysis, one study reported that adults with HBW had lower DBP than those with NBW (statistical test unavailable).47 The other study revealed that DBP was higher in children with HBW than those with NBW (not a statistically significant difference)23 (Figure 3).

Hypertension

Fifteen studies with 31 RRs for hypertension associated with HBW and NBW were included. The overall RR combination was 1.00 (95% confidence interval, 0.931, 1.06) with significant heterogeneity (I2=63.8%, Q-statistic: P<0.0001). The subgroup analysis showed that there was heterogeneity in the age group >40 years. With increasing age, the RR for hypertension changed from >1 to <1. The results indicate that HBW is associated with a higher risk of hypertension in younger subjects but a lower risk in older subjects (Figure 4).

Figure 4
figure 4

Risk ratio for hypertension between HBW and NBW in meta-analysis. Studies ordered by the mean age at which the event was checked. The pooled risk ratios were calculated by means of a random-effects model; 95% confidence intervals (CI) are shown in parentheses and as horizontal bars. HBW, high birth weight, NBW, normal birth weight, RR, risk ratio.

Full size image

The sources of heterogeneity were investigated by performing meta-regression analysis. The results of the meta-regression show that age was the primary, statistically significant source of heterogeneity (Table 5). Age was inversely associated with the RR of hypertension, indicating that older subjects with HBW had a lower risk of hypertension than those with NBW (Figure 4).

Table 5 Meta-regression for the RR of hypertension between HBW and NBW
Full size table

Discussion

In this review, we primarily aimed to assess the effects of HBW on blood pressure and hypertension by summarizing the current evidence from published studies. However, substantial heterogeneity made it inappropriate to synthesize all of the data. In particular, the subgroup analysis showed that some subgroups exhibited high heterogeneity. Further meta-regression analysis revealed that age was consistently associated with the effects of HBW on blood pressure and the risk of hypertension.

The results of both the meta-combination and meta-regression analysis revealed that age was inversely associated with the effects of HBW on blood pressure and hypertension. The results indicate that HBW has contrasting effects on blood pressure and hypertension in younger subjects compared with older subjects. In the younger subjects, HBW resulted in higher blood pressure and a higher risk of hypertension than NBW. In older subjects, HBW led to lower blood pressure and a lower risk of hypertension.

HBW, which indicates overnutrition in the initial stage of life, raised blood pressure and the risk of hypertension in early life. Another published systematic review also reported that newborns with higher birth weight had higher blood pressure.16 Although an age-related increase in blood pressure has been observed in almost every population,53 the increase in blood pressure associated with HBW seems to be attenuated or reversed with increasing age. In other words, a ‘catch-down’ effect occurs in blood pressure and the risk of hypertension when individuals with HBW grow older.

In fact, a ‘catch-down’ phenomenon is common in babies born large. For example, large for gestational age babies usually experience a postnatal ‘catch-down’ in height, weight and so on.54, 55 The increase in blood pressure likely follows the same pattern for weight or height in large babies, with this increase subsiding after several years.

Evidence has shown that subjects with HBW are usually bigger in size and heavier in weight.56 However, these subjects are more metabolically healthy when they grow older because they have more lean mass than adipose tissue.57 This condition may lead to lower blood pressure in subjects with HBW compared with those who have the same BMI but were born with NBW. Another possible explanation is that the difference in blood pressure between those with HBW and NBW is very subtle in childhood, and other factors compensate for these differences later in life.

To the best of our knowledge, this review is the first to discuss the association of HBW with blood pressure and hypertension over the life course. This review reveals that age is an important factor that is associated with birth weight and affects blood pressure and hypertension. Considering the conflicting implications of HBW (that is, an overweight status in the early stage of life does not correspond to a high risk of hypertension in later stages of life), we propose that the ‘catch-down’ effect in blood pressure might attenuate the risk of hypertension associated with HBW.

However, this review has some limitations. First, only studies published in English were eligible for the study, which may have introduced selection bias. Second, given that age is the main source of heterogeneity, subgroup meta-analyses of more specific age groups may be more effective when additional evidence is available. Third, subjects with HBW are a heterogeneous population that includes both individuals from diabetic pregnancies or obese mothers and normal large infants with birth weights that align with their growth potential. Considering that information on the etiology of fetal overgrowth is not reported in most studies, the effects of specific causes of HBW on blood pressure remain unclear.

Conclusion

The MDs in blood pressure and in the RR of hypertension between individuals with HBW and NBW are inversely associated with age. Blood pressure and the risk of hypertension are higher among individuals with HBW during childhood but lower during adulthood. This finding may be partially attributed to the ‘catch-down’ effect in the elevation of blood pressure when the subjects with HBW grow older.