Maternal haemoglobin levels and cardio-metabolic risk factors in childhood: the Generation R Study

Authors

  • M Welten,

    1. The Generation R Study Group, Erasmus Medical Center, Rotterdam, the Netherlands
    Search for more papers by this author
  • R Gaillard,

    1. The Generation R Study Group, Erasmus Medical Center, Rotterdam, the Netherlands
    2. Department of Paediatrics, Erasmus Medical Center, Rotterdam, the Netherlands
    Search for more papers by this author
  • A Hofman,

    1. Department of Epidemiology, Erasmus Medical Center, Rotterdam, the Netherlands
    Search for more papers by this author
  • LL de Jonge,

    1. The Generation R Study Group, Erasmus Medical Center, Rotterdam, the Netherlands
    2. Department of Paediatrics, Erasmus Medical Center, Rotterdam, the Netherlands
    Search for more papers by this author
  • VWV Jaddoe

    Corresponding author
    1. The Generation R Study Group, Erasmus Medical Center, Rotterdam, the Netherlands
    2. Department of Paediatrics, Erasmus Medical Center, Rotterdam, the Netherlands
    • Correspondence: VWV Jaddoe, MD PhD, The Generation R Study Group (Na29-15), Erasmus Medical Center, PO Box 2040, 3000 CA Rotterdam, The Netherlands. Email v.jaddoe@erasmusmc.nl

    Search for more papers by this author

Abstract

Objective

To assess whether variations in maternal haemoglobin levels during pregnancy are associated with cardio-metabolic risk factors in school age children.

Design

Population-based prospective cohort study.

Setting

Rotterdam, The Netherlands, 2002–2012.

Population

Mothers and children (= 5002) participating in the Generation R Study.

Methods

We obtained maternal haemoglobin levels during early pregnancy (median gestational age 14.6 weeks [95% range 10.3, 25.3]) from venous blood samples. Maternal anaemia and elevated haemoglobin levels were based on World Health Organization criteria. We measured childhood cardio-metabolic risk factors at age 6 years.

Main outcome measures

Cardio-metabolic risk factors included body mass index, total fat mass percentage, android/gynoid fat mass ratio, systolic and diastolic blood pressure, left ventricular mass, and blood levels of cholesterol, insulin and C-peptide.

Results

Maternal haemoglobin levels were not associated with childhood body mass index, total fat mass percentage, android/gynoid fat mass ratio, systolic blood pressure, cholesterol or insulin levels. Compared with children with normal maternal haemoglobin levels, children from anaemic mothers had slightly higher diastolic blood pressures (difference 0.70 mmHg, 95% CI 0.12, 1.29) and lower C-peptide levels (difference factor 0.93, 95% CI 0.88, 0.98), and children of mothers with elevated haemoglobin levels had lower left ventricular masses (difference −1.08 g, 95% CI −1.88, −0.29). These associations attenuated after adjustment for multiple testing and were not consistent within linear models.

Conclusion

These results do not strongly support the hypothesis that variations in maternal haemoglobin levels during pregnancy influence cardio-metabolic risk factors in childhood.

Introduction

Maternal anaemia and elevated haemoglobin levels are related to adverse pregnancy outcomes.[1-3] Both maternal anaemia and elevated haemoglobin levels may affect the fetal supply line and cause a restricted intrauterine environment.[4, 5] An accumulating body of evidence suggests that fetal adaptive responses in response to a restricted intrauterine environment lead to cardiovascular disease and type 2 diabetes in later life.[6] Previous studies have demonstrated associations of low birthweight with cardiovascular disease, type 2 diabetes and their risk factors in childhood and adulthood.[7] Maternal anaemia and elevated haemoglobin levels during pregnancy are associated with low birthweight.[1, 3, 8] Whether and to what extent low and high maternal haemoglobin levels during pregnancy also affect cardio-metabolic risk factors and disease in later life is largely unknown. Studies in rats have shown associations of maternal anaemia during pregnancy with increased systolic blood pressure and heart-to-body weight in the offspring.[9-12] In humans, a positive association of maternal haemoglobin levels during pregnancy with systolic blood pressure in children aged 5–9 years has been reported.[13] However, these results are not consistent.[14-17] Also, previous studies on the associations of maternal haemoglobin levels with children's cardiovascular risk factors were mainly focused on blood pressure. Not much is known about the associations of maternal haemoglobin levels during pregnancy with other, more detailed, childhood cardio-metabolic risk factors.

Therefore, in a population-based prospective cohort study among 5002 mothers and their children, we assessed the associations of maternal haemoglobin levels in pregnancy with cardio-metabolic risk factors in 6-year-old children.

Methods

Design and study population

The present study was embedded in the Generation R Study, a population-based prospective cohort study from early pregnancy onwards in Rotterdam, the second largest city in The Netherlands.[18-20] The study was designed to identify early environmental and genetic causes and causal pathways leading to normal and abnormal growth, development, and health during fetal life, childhood, and adulthood, and has been described in detail previously.[18] Pregnant women resident in the study area at time of delivery and with a delivery date between April 2002 and January 2006 were considered eligible.[18] The study aimed to enrol women early in pregnancy, but enrolment was allowed until birth of the child.[18] Of the 8633 prenatally enrolled singleton live-born children, 7312 children had information on maternal haemoglobin levels. Missing haemoglobin levels were mainly due to unavailability of the routine care data from midwife and obstetric registries. A total of 5033 (69%) children visited the research centre with their parents for follow-up measurements for the present study (median age 6.0 years, 95% range 5.7, 7.8). Children with echographic evidence of heart disease or kidney disease were excluded from the study (= 31), leaving 5002 children for the current analyses (Figure 1). The study has been approved by the Medical Ethics Committee of the Erasmus Medical Centre, Rotterdam. Written informed consent was obtained from all parents.[21]

Figure 1.

Flow chart describing the selection of the population for analysis.

Maternal haemoglobin concentrations

Maternal venous blood samples were collected in early pregnancy.[22] Median gestational age at blood sampling was 14.6 weeks (95% range 10.3, 25.3). Haemoglobin concentrations were measured in fresh ethylene diamine tetra-acetic acid (EDTA) plasma samples. We constructed gestational age adjusted standard deviation scores (SDS) for haemoglobin levels using a model that included a spline of the trend of the mean with a chance distribution independent from gestational age (see Appendix S1 for details). We subsequently used these SDS as continuous variables and as quintiles to assess linear and non-linear associations. We also used clinical cut-offs and defined maternal anaemia and elevated haemoglobin levels as the 12.5% lowest and highest haemoglobin levels, which corresponds with the World Health Organization (WHO) criteria for anaemia in this cohort. The WHO defines anaemia in pregnant women as haemoglobin concentrations <11.00 g/dl (6.83 mmol/l)[23], which reflected the lowest 12.5% in our study population. Similarly, levels equal to and higher than the 87.5th percentile, reflecting a haemoglobin concentration ≥13.21 g/dl (8.20 mmol/l), were considered elevated haemoglobin levels. We performed sensitivity analyses using the lower and upper 5% of maternal haemoglobin levels as a more stringent cut off for anaemia (≤10.47 g/dl or 6.50 mmol/l) and elevated haemoglobin levels (≥13.70 g/dl or 8.50 mmol/l).

Childhood outcomes

Weight and height of the child were assessed without shoes or heavy clothing at the age of 6 years. Child body mass index (BMI) was calculated as the ratio of the weight of the child in kilograms and the squared height of the child in meters.

Body composition measurements of children were obtained by a Dual-Energy X-ray absorptiometry (DXA) scan (iDXA; General Electrics–Lunar, 2008, Madison, WI, USA). All scans were performed with the same device and software (enCORE). Total fat mass was expressed as a percentage of total body weight and android/gynoid fat mass ratio was calculated.

We measured systolic and diastolic blood pressure four times at 1-minute intervals at the right brachial artery using the validated automatic sphygmanometer Datascope Accutor Plus (Paramus, NJ, USA).[24] A cuff with a width of approximately 40% of the arm circumference and long enough to cover 90% of the arm circumference was selected. Four successful blood pressure measurements were available for over 90% of the children.[19, 20]

Two-dimensional M-mode echocardiographic measurements were performed to measure interventricular end-diastolic septal thickness (IVSTD), left ventricular end-diastolic diameter (LVEDD), and left ventricular end-diastolic posterior wall thickness (LVPWTD) using methods recommended by the American Society of Echocardiography.[25] Left ventricular mass (LV mass) was calculated with the formula derived by Devereux et al.[26]

Thirty minutes’ fasting blood samples were collected in 69% of the children by ante-cubital venepuncture. Missing values were mainly due to no consent or crying of the child. We measured total-, high density lipoprotein (HDL)-, and low density lipoprotein (LDL)-cholesterol, insulin and C-peptide concentrations enzymatically using a Cobas 8000 analyser (Roche, Almere, The Netherlands). Quality control samples demonstrated intra- and interassay coefficients of variation ranging from 0.77 to 1.39%, and from 0.87 to 2.40%, respectively.

Covariates

Information on maternal age, pre-pregnancy weight, parity, educational level, smoking status during pregnancy, alcohol intake during pregnancy, folic acid intake, and ever having breastfed the child was obtained by questionnaires. Maternal height was measured without shoes and pre-pregnancy BMI was calculated (kg/m2). Maternal blood pressure was measured at enrollment, which was the same gestational age at which the haemoglobin blood sample was collected, using the Omron 907® automated digital oscillometric sphygmomanometer (OMRON Healthcare Europe B.V., Hoofddorp, The Netherlands). Date of birth, gestational age at birth, birthweight and child's sex were obtained from midwife and hospital registries.[19, 20]

Statistical analysis

We conducted non-response analysis to assess differences in maternal and child characteristics between those participating and not participating in the follow-up studies, using t-tests, Mann–Whitney U-tests, and Chi-square tests.

We performed linear regression analyses to assess the associations of maternal haemoglobin levels (continuous SDS; SDS quintiles; and clinical cut offs (anaemia, normal haemoglobin levels, and elevated haemoglobin levels)) with childhood BMI, body fat distribution, blood pressure, left ventricular mass and levels of cholesterol, insulin and C-peptide. As blood pressure was measured four times, we used linear mixed models, which fit the four blood pressure measurements per child as repeated outcome measures.[27] Compared with using the mean of four blood pressure measurements, this approach has the advantage of assigning higher weights in analysis to subjects with the highest number of measurements available and the least individual variability in their blood pressure measurements.[28, 29] We transformed body fat distribution, insulin and C-peptide outcomes with the natural logarithm because of their positively skewed distributions. The regression coefficients of the log-transformed outcome data were transformed back by taking the natural antilogarithm e, the base of the natural logarithm, of these regression coefficients. First, we adjusted the regression models for child's sex and current age only. Models for the body fat distributions were additionally adjusted for child height and models for left ventricular mass for ultrasound device and performing sonographer. The analyses of anaemia and elevated haemoglobin levels were adjusted for gestational age at the time of the haemoglobin level assessment. These models were considered the basic models. We also adjusted these basic models for potential confounders including maternal age, pre-pregnancy BMI, ethnicity, educational level, parity, blood pressure, smoking, alcohol consumption, folic acid intake, breastfeeding of the child, gestational age at birth, and birthweight of the child. Moreover, child BMI was a covariate in the analyses of blood pressure, left ventricular mass and levels of cholesterol, insulin and C-peptide. These analyses were considered cofounder models. All covariates had <20% missing values, except maternal folic acid use (24%). Missing values were imputed with multiple imputation techniques (= five imputations). The imputed datasets were analysed together. Effect modification by child's sex was tested in the fully adjusted models, as previous studies suggested, but no consistent evidence of interaction was found in the present study (data not shown). We applied Bonferroni correction to adjust for multiple testing. Because we analysed three groups of outcome variables, a P-value of 0.017 instead of 0.05 was used. All measures of association are presented with their 95% confidence intervals (CI). Analyses were performed using the Statistical Package for the Social Sciences version 20.0 for Windows (SPSS Inc., Chicago, IL, USA). The mixed-model analyses were performed using Statistical Analysis System version 9.2 (SAS Institute Inc., Gary, NC, USA).

Results

Table 1 shows the characteristics of the study population after imputation. Table S1 shows the non-imputed characteristics of the study population. According to the WHO criteria, 624 mothers were considered anaemic based on their haemoglobin levels (median haemoglobin level 10.63 g/dl, 95% range 8.70, 10.96). Elevated haemoglobin levels were found in 730 mothers (median haemoglobin level 13.54 g/dl, 95% range 13.21, 14.66), leaving 3648 mothers with normal haemoglobin levels (median haemoglobin level 12.09 g/dl, 95% range 11.12, 13.05). The main results of the non-response analyses indicated that compared with mothers who were excluded because of missing data on child cardio-metabolic risk factors, mothers included in the analyses were less likely to be anaemic and more likely to be older, and have European ethnicity, higher educational level, and offspring with higher weight and gestational age at birth (Table S2). The results from the basic models, adjusted for sex and age only, are given in the supplementary materials (Tables S3–S5). The results of the confounder models, i.e. main models, are shown in Tables 2-4.

Table 1. Characteristics of the study population
 = 5002
  1. Values are expressed as mean (SD), median (95% range) or number (%).

  2. Missing values on covariates were imputed with multiple imputation techniques.

Maternal characteristics
Age (years)30.9 (19.8–39.4)
Height (cm)167.5 (7.4)
Weight before pregnancy (kg)66.4 (12.6)
Body mass index before pregnancy (kg/m2)23.6 (4.3)
Systolic blood pressure (mmHg)115.9 (12.2)
Diastolic blood pressure (mmHg)68.0 (9.6)
Parity (%)
02855.2 (57.1)
≥12146.8 (42.9)
Ethnicity (%)
European2984.0 (59.7)
Non-European2018.0 (40.3)
Educational level (%)
No education or primary538.0 (10.8)
Secondary2211.2 (44.2)
Higher2252.8 (45.0)
Smoking (%)
Never3705.0 (74.1)
Until pregnancy was known444.6 (8.9)
Yes, continued during pregnancy852.4 (17.0)
Alcohol consumption (%)
Never2354.6 (47.1)
Until pregnancy was known682.8 (13.7)
Yes, continued during pregnancy1964.6 (39.3)
Folic acid supplement use (%)
Never1367.6 (27.3)
Started within first 10 weeks of pregnancy1541.2 (30.8)
Periconception2093.2 (41.8)
Breastfeeding (%)
No376.4 (7.5)
Yes4625.6 (92.5)
Haemoglobin (g/dl)12.1 (1.0)
Haemoglobin clinical categories (%)
Anaemia (<11.00 g/dl)624 (12.5)
Normal (11.00–13.21 g/dl)3648 (72.9)
Elevated (≥13.21 g/dl)730 (14.6)
Child characteristics
Male sex (%)2481 (49.6)
Gestational age at birth (weeks)40.1 (35.9–42.3)
Birth weight (g)3428 (550)
Age at follow up (years)6.0 (5.7–7.8)
Height (cm)119.2 (5.9)
Weight (kg)23.2 (4.2)
Body mass index (kg/m2)16.2 (1.9)
Total fat mass percentage24.0 (16.3–39.1)
Android/gynoid ratio0.2 (0.2–0.4)
Systolic blood pressure (mmHg)102.6 (8.1)
Diastolic blood pressure (mmHg)60.6 (6.8)
Left ventricular mass (g)53.2 (11.3)
Total cholesterol (mmol/l)4.2 (0.6)
HDL cholesterol (mmol/l)1.4 (0.3)
LDL cholesterol (mmol/l)2.4 (0.6)
Insulin (pmol/l)113.2 (16.5–397.9)
C-peptide (nmol/l)1.0 (0.3–2.1)
Table 2. Associations of maternal haemoglobin levels with childhood body mass index and body fat distribution
 Body mass index (kg/m2) = 4995Total fat mass percentage = 4885Android/gynoid ratio = 4885
  1. Values are regression coefficients (95% confidence intervals) and reflect the change in BMI and body fat distribution per change of 1 SDS or haemoglobin category. Models are adjusted for the child factors gestational age, birthweight, sex and current age and the maternal factors age, pre-pregnancy BMI, ethnicity, educational level, parity, blood pressure, smoking, alcohol consumption, folic acid intake and breastfeeding. The fat distribution measurements were log-transformed and additionally adjusted for child's height.

Haemoglobin (SDS)

0.01

(−0.05, 0.06)

1.01

(1.00, 1.01)

1.00

(1.00, 1.01)

Haemoglobin quintiles

SDS ≤ −0.80

= 1001

−0.01

(−0.17, 0.15)

0.99

(0.97, 1.00)

0.99

(0.97, 1.01)

SDS −0.80, −0.24

= 1005

−0.11

(−0.26, 0.05)

0.99

(0.97, 1.00)

1.00

(0.98, 1.02)

SDS −0.24, 0.29

= 993

ReferenceReferenceReference

SDS 0.29, 0.85

= 1003

−0.02

(−0.17, 0.14)

1.00

(0.99, 1.02)

1.00

(0.98, 1.02)

SDS > 0.85

= 1000

−0.04

(−0.19, 0.11)

0.99

(0.97, 1.01)

1.00

(0.98, 1.02)

Haemoglobin clinical categories

Anaemia

= 624

−0.01

(−0.17, 0.14)

0.99

(0.97, 1.00)

0.99

(0.97, 1.01)

Normal

= 3648

ReferenceReferenceReference

Elevated

= 730

0.07

(−0.07, 0.21)

1.00

(0.99, 1.02)

1.00

(0.98, 1.02)

Table 3. Associations of maternal haemoglobin levels with childhood blood pressure and left ventricular mass
 Systolic blood pressure (mmHg) = 4843Diastolic blood pressure (mmHg) = 4843Left ventricular mass (g) = 4630
  1. Values are regression coefficients (95% CI) and reflect the change in blood pressure and left ventricular mass per change of 1 SDS or haemoglobin category. Models are adjusted for the child factors gestational age, birthweight, sex, current age, and BMI and the maternal factors age, pre-pregnancy BMI, ethnicity, educational level, parity, blood pressure, smoking, alcohol consumption, folic acid intake and breastfeeding. Models with the outcome left ventricular mass are additionally adjusted for ultrasound device and performing sonographer.

  2. *P < 0.05; **P < 0.01.

Haemoglobin (SDS)

−0.11

(−0.34, 0.12)

−0.27

(−0.46, −0.07)**

−0.35

(−0.63, −0.06)*

Haemoglobin quintiles

SDS ≤ −0.80

= 1001

0.01

(−0.67, 0.68)

0.12

(−0.46, 0.70)

0.76

(−0.10, 1.61)

SDS −0.80, −0.24

= 1005

0.21

(−0.46, 0.87)

−0.05

(−0.63, 0.52)

0.73

(−0.12, 1.57)

SDS −0.24, 0.29

= 993

ReferenceReferenceReference

SDS 0.29, 0.85

= 1003

−0.25

(−0.91, 0.42)

−0.46

(−1.03, 0.11)

0.31

(−0.53, 1.15)

SDS > 0.85

= 1000

−0.15

(−0.82, 0.52)

−0.58

(−1.16, −0.01)*

−0.26

(−1.12, 0.59)

Haemoglobin clinical categories

Anaemia

= 624

0.27

(−0.41, 0.96)

0.70

(0.12, 1.29)*

0.26

(−0.60, 1.11)

Normal

= 3648

ReferenceReferenceReference

Elevated

= 730

−0.04

(−0.67, 0.59)

−0.20

(−0.74, 0.34)

−1.08

(−1.88, −0.29)**

Table 4. Associations of maternal haemoglobin levels with childhood cholesterol, insulin and C-peptide levels
 Total cholesterol (mmol/l) = 3311HDL cholesterol (mmol/l) = 3314LDL cholesterol (mmol/l) = 3313Insulin (pmol/l) = 3287C-peptide (nmol/l) = 3290
  1. Values are regression coefficients (95% CI) and reflect the change in levels of cholesterol, insulin, and C-peptide per change of 1 SDS or haemoglobin category. Models are adjusted for the child factors gestational age, birthweight, sex, current age, and BMI and the maternal factors age, pre-pregnancy BMI, ethnicity, educational level, parity, blood pressure, smoking, alcohol consumption, folic acid intake and breastfeeding. The outcomes insulin and C-peptide level were log-transformed.

  2. **P < 0.01.

Haemoglobin (SDS)

−0.02

(−0.04, 0.00)

−0.01

(−0.02, 0.00)

−0.01

(−0.03, 0.01)

1.01

(0.98, 1.04)

1.01

(1.00, 1.03)

Haemoglobin quintiles

SDS ≤ −0.80

= 1001

0.03

(−0.04, 0.10)

0.01

(−0.02, 0.05)

0.04

(−0.03, 0.10)

1.00

(0.91, 1.09)

0.97

(0.92, 1.02)

SDS −0.80, −0.24

= 1005

0.01

(−0.06, 0.08)

−0.02

(−0.06, 0.01)

0.01

(−0.05, 0.07)

1.00

(0.91, 1.09)

0.98

(0.93, 1.03)

SDS −0.24, 0.29

= 993

ReferenceReferenceReferenceReferenceReference

SDS 0.29, 0.85

= 1003

0.01

(−0.05, 0.08)

−0.02

(−0.05, 0.02)

0.03

(−0.03, 0.09)

1.00

(0.91, 1.09)

0.98

(0.93, 1.04)

SDS > 0.85

= 1000

−0.03

(−0.10, 0.04)

−0.02

(−0.06, 0.01)

0.00

(−0.06, 0.07)

1.01

(0.92, 1.10)

1.00

(0.95, 1.05)

Haemoglobin clinical categories

Anaemia

= 624

0.03

(−0.04, 0.10)

0.03

(0.00, 0.07)

0.04

(−0.02, 0.10)

0.92

(0.84, 1.00)

0.93

(0.88, 0.98)**

Normal

= 3648

ReferenceReferenceReferenceReferenceReference

Elevated

= 730

−0.01

(−0.07, 0.06)

0.01

(−0.02, 0.04)

0.01

(−0.05, 0.07)

1.03

(0.95, 1.12)

1.02

(0.97, 1.07)

We did not observe any associations of maternal haemoglobin levels with childhood BMI, total fat mass percentage or android/gynoid fat mass ratio (Table 2).

Similarly, no associations were found for maternal haemoglobin levels with childhood systolic blood pressure (Table 3). We observed a trend for an inverse association of maternal haemoglobin levels with childhood diastolic blood pressure (P for trend <0.01). For the fifth SDS quintile of maternal haemoglobin levels we found a 0.58 mmHg (95% CI −1.16, 0.01) decrease in diastolic blood pressure compared with the third quintile. No associations were found for the other SDS quintiles or for elevated maternal haemoglobin levels with diastolic blood pressure. Children whose mothers were anaemic during pregnancy had a 0.70 mmHg (95% CI 0.12, 1.29) higher diastolic blood pressure, as compared with children whose mothers had haemoglobin levels within the normal range. Also, we observed a trend for an inverse association of maternal haemoglobin levels with childhood left ventricular mass (P for trend <0.05). No associations for SDS quintiles or maternal anaemia with left ventricular mass were found. Children whose mothers had elevated haemoglobin levels during pregnancy had a 1.08 g (95% CI −1.88, −0.29) lower left ventricular mass compared with children whose mothers had haemoglobin levels within the normal range. After adjustment for multiple testing, the association for maternal anaemia with diastolic blood pressure attenuated to non-significant.

The fully adjusted models for the associations of maternal haemoglobin levels with cholesterol, insulin and C-peptide levels are shown in Table 4. No associations were observed between maternal haemoglobin levels and children's cholesterol or insulin levels. Children whose mothers were anaemic during pregnancy had lower C-peptide levels by a factor of 0.93 (95% CI 0.88, 0.98) compared with children whose mothers had haemoglobin levels within the normal range, but we did not observe a trend or any associations for SDS quintiles or elevated maternal haemoglobin levels with levels of C-peptide. This association remained significant after adjustment for multiple testing.

Sensitivity analyses using the more stringent cut off of the lower and upper 5% of haemoglobin levels for severe anaemia (median haemoglobin level 10.15 g/dl, 95% range 8.06, 10.47) and elevated haemoglobin levels (median haemoglobin level 13.86 g/dl, 95% range 13.70, 14.99) showed similar results as the analysis defining anaemia and elevated haemoglobin levels using the WHO criteria (Table S6).

Discussion

Summary of main findings

We did not observe any associations between maternal haemoglobin levels and childhood BMI, body fat distribution, systolic blood pressure, and levels of cholesterol or insulin. Maternal anaemia was associated with a minimally increased childhood diastolic blood pressure and lower C-peptide levels. Maternal elevated haemoglobin level was associated with lower left ventricular mass in offspring. The association of maternal anaemia with diastolic blood pressure attenuated to non-significant after taking multiple testing into account. The results from this study are important from an aetiological perspective. We acknowledge that the observed effect estimates were small and reflect subclinical changes in cardiovascular and metabolic function in school age children. None of the children had known cardiovascular disease. Further follow-up studies are needed to explore whether maternal haemoglobin levels lead to cardiovascular and metabolic disease in adulthood.

Strengths and limitations of the study

The main strength of this study is its population-based prospective design, which started from fetal life onwards. At baseline, the response rate of the study cohort was 61%. Extensive information on pregnancy, and maternal and child characteristics was obtained from participants. Maternal haemoglobin levels were only measured in mothers who were enrolled in the first or second trimester of pregnancy and were therefore not available for the full cohort. Follow-up response rate was 69%. Non-response analysis did indicate some differences between persons included in the analyses and those lost to follow-up. These differences would have led to selection bias if the associations of maternal haemoglobin levels with cardio-metabolic risk factors had been different for the two populations. This seems unlikely, but cannot be excluded. We used different approaches to explore the influence of maternal haemoglobin levels on cardiovascular and metabolic development in school-age children. We also explored the role of both low and elevated haemoglobin levels, as both extremes seem to be associated with increased risks of adverse birth outcomes.[5, 30] We had no information on changes in haemoglobin level during pregnancy, as only a single measurement was available in the full population for analysis. Unfortunately, we also did not have information available on either maternal or child haemoglobinopathies or on the child's haemoglobin level at age 6 years and were therefore unable to take this information into account. We had information on many potential confounders. Despite extensive data collection we did not have any information available on maternal iron (supplement) intake throughout pregnancy. Women may have received iron supplementation after detection of low haemoglobin levels, possibly making it harder to observe potential associations.[15] We were not able to assess these possible effects of iron supplementation during pregnancy.

A limitation of our study was that we only had 30-minute fasting blood samples available. As children were invited to visit the research centre throughout the day, it was not feasible to request the children to fast. Because of this short fasting period and the high variability in time of day of visitation it is possible that differences in insulin and C-peptide levels between individuals were mainly due to the differences in the food content ingested by the children. However, as we are looking into the association between maternal haemoglobin levels and child's cardio-metabolic outcomes, we do not expect this to have had a great impact on the results, as timing of blood drawn and nutritional intake of the children beforehand was not related to maternal haemoglobin levels. Moreover, for the cholesterol levels a 30-minute fasting blood sample is not likely to affect our results, as a recent study concluded that fasting is not really necessary for routine lipid level determination.[31]

Comparison with the existing literature

In pregnancy, normal haemodynamic changes lead to a greater increase in plasma volume compared with the increase in red blood cells leading to physiologic anaemia.[32] In contrast to iron deficiency anaemia, physiologic anaemia is thought to be associated with better outcomes for mother and child.[1, 5, 32, 33] Reduced blood viscosity can lead to better perfusion of the placenta and a lower maternal cardiac work load.[32] Secondly, the increased blood volume forms a reserve for the blood lost during parturition.[32] Because plasma volume expansion and iron deficiency lead to lower haemoglobin concentrations, both women with physiologic and iron deficiency anaemia are considered anaemic despite differences in aetiology and consequences.

The effects of maternal haemoglobin levels on pregnancy outcomes have been explored extensively.[1-3, 34, 35] A previous analysis from the same cohort as the present study, showed that elevated maternal haemoglobin levels, and not anaemia, were associated with fetal head circumference, length, and weight growth restriction from the third trimester onwards compared with normal maternal haemoglobin levels.[36] Elevated haemoglobin levels were also associated with increased risks of gestational hypertensive disorders and adverse birth outcomes.[36] Less is known about the long-term consequences for the children. We hypothesised that both maternal anaemia and elevated haemoglobin levels may cause restricted intrauterine environments[4, 5] and subsequently evoke cardiovascular and metabolic adaptive responses in the fetus, which may lead to cardiovascular disease and type 2 diabetes in later life. To our knowledge, this is the first study focused on the associations of maternal haemoglobin levels during pregnancy with multiple detailed measured childhood cardio-metabolic risk factors. Moreover, with a sample size of over 5000 mothers and children, this study population is the largest in which the effects of maternal haemoglobin on childhood blood pressure have been assessed. Various studies in both animals and humans have previously explored associations of maternal haemoglobin status during pregnancy with cardiovascular and metabolic risk factors in the offspring.[9-17, 37-41]

Komolova et al.[37] observed significantly increased visceral adipose tissue expressed as body weight percentage in perinatal iron-deficient rats compared with controls. In the present study, we observed no direct association of maternal anaemia during pregnancy with children's BMI or body fat distribution. A possible explanation for the difference in findings might be that in the animal study, the induced iron deficiency was more severe than the observed anaemia and natural variation in haemoglobin levels in a well-nourished human population. We are not aware of other studies exploring these associations in children.

Also, several studies in rats observed higher systolic blood pressure in offspring of iron-deficient dams than their controls.[9-12] Two studies indicated an initially lower but later higher systolic blood pressure in offspring of iron-deficient dams.[9, 10] Most studies performed in humans have not found an association of maternal anaemia or haemoglobin levels with systolic blood pressure in children after adjustment for confounders.[14-17] Although Bergel et al.[13] observed a positive association between continuous maternal haemoglobin levels and systolic blood pressure in 518 Argentinean children aged 5–9 years, no association with maternal anaemia was found. In another prospective cohort consisting of 1255 children aged 7 years from the UK, Brion et al.[15] observed an association of early pregnancy anaemia with lowered diastolic blood pressures in children, and with lower systolic blood pressure of children for a subgroup of women who did not take iron supplements. An explanation for the differences in results between animal and human studies might be that maternal anaemia in rats induced by iron deficiency is more severe than the observed anaemia in humans from unspecified causes. We observed no associations of maternal haemoglobin levels during pregnancy with children's systolic blood pressure. Our results indicated that maternal anaemia was associated with a slightly increased diastolic blood pressure but overall results were inconsistent and this association might be a chance finding due to the multiple testing. We cannot explain what might have caused this difference from the results of Brion et al. A possible explanation might be that only very severely decreased or elevated maternal haemoglobin levels during pregnancy have an effect on the children's blood pressure and that there is no effect of variations within the normal range of haemoglobin levels. In this case, it is likely that population-based observational studies do not include enough severe cases of maternal elevated and decreased haemoglobin values to detect any effects. Also, previously reported associations of maternal haemoglobin levels or anaemia with children's blood pressure in observational studies may be explained by residual confounding.

Studies in both rat and sheep found that maternal iron deficiency and anaemia led to higher heart to body weights, but results were not consistent.[4, 9-11, 38, 39] It has been hypothesised that maternal anaemia leads to larger hearts in offspring because of cardiovascular adaptations to maintain adequate oxygen supply under circumstances of reduced oxygen content by increasing cardiac output and stroke volume.[4] We are not aware of any human studies focusing on the association of maternal anaemia with heart size. In this study, we did observe an association of maternal elevated haemoglobin levels with lower left ventricle mass adjusted for childhood BMI. As no other studies are available, our results should be interpreted cautiously and need replication in humans.

Studies in rats found no differences in levels of cholesterol[11, 12, 40, 41] or insulin[10-12] between the offspring of anaemic or iron-deficient dams and their controls. Gambling et al.[10] did not find a difference in glucose tolerance among 10-week old offspring of iron-deficient dams and their controls. Lewis et al.[11, 12] found that offspring of iron-deficient dams had a better glucose tolerance than their controls at age 3 months but had similar glucose tolerance at age 13 months. To our knowledge the present study is the first to explore the associations of maternal haemoglobin levels during pregnancy with offspring's cholesterol and insulin levels in humans. In line with studies in animals, we found no consistent associations for maternal anaemia during pregnancy and cholesterol or insulin levels at the age of 6 years. However, we did observe an association for maternal anaemia and lower C-peptide levels in children. We are not aware of any other studies looking into the association between maternal haemoglobin and child C-peptide levels in either animals or humans.

Conclusion

The results of this study do not strongly support the hypothesis that variations in maternal haemoglobin levels during pregnancy, in either clinical extremes or across the normal range, influence cardiovascular or metabolic outcomes in school-age children. Whether the observed associations of maternal haemoglobin levels with left ventricular mass and C-peptide levels have long-term consequences remains a topic for further research.

Disclosure of interests

The authors do not have any potential conflicts of interest to disclose.

Contribution to authorship

MW and VWVJ designed the study. MW analysed the data and wrote the manuscript. RG, AH, LLdJ and VWVJ provided comments and consultation on the analyses. RG, AH, LLdJ and VWVJ provided comments and consultation on the manuscript. All authors read and approved the final manuscript.

Details of ethics approval

The Medical Ethics Committee at Erasmus University Medical Centre, Rotterdam, The Netherlands, approved the study (MEC 198.782/2001/31). Written informed consents were obtained from all participants in the study.

Funding

The first phase of the Generation R Study has been made possible by financial support from the Erasmus Medical Centre, Rotterdam; the Erasmus University Rotterdam; and The Netherlands Organization for Health Research and Development. Vincent Jaddoe received an additional grant from The Netherlands Organization for Scientific Research (NWO VIDI). Additional support was provided by the Dutch Heart Foundation (2008B114).

Acknowledgements

The Generation R study is being conducted by the Erasmus Medical Centre in close collaboration with the School of Law and Faculty of Social Sciences of Erasmus University Rotterdam; the Municipal Health Service Rotterdam area, Rotterdam; the Rotterdam Homecare Foundation, Rotterdam; and the Stichting Trombosedienst & Artsenlaboratorium Rijnmond (STAR), Rotterdam. We gratefully acknowledge the contributions of general practitioners, hospitals, midwives and pharmacies in Rotterdam.

Ancillary