Search strategy

Comprehensive literature searches were independently performed in May 2015 by a team of researchers. Although this is not a generic systematic review, this paper followed a systematic approach to select all available studies describing MNS and how it associates with fetal, birth, neonatal and/or infant outcomes in Africa. Database used to conduct the searches included the following: Medline, EMBASE, Web of Science, Google Scholar, ScienceDirect, SciSearch and Cochrane Library. Search terms and phrases included the following, as well as variations of the following where applicable: prenatal/anthropometry/(specific anthropometric measure of interest, e.g. body mass index)/maternal nutrition/(specific nutrient of interest, e.g. protein or iron)/(specific micronutrient deficiency of interest, e.g. anaemia)/(specific nutritional biomarker of interest, e.g. ferritin)/(specific dietary intake assessment method of interest, e.g. food frequency questionnaire) and birth outcome/pregnancy outcome/(specific adverse outcome of interest, e.g. low birth weight)/(specific growth or body composition variable of interest, e.g. head circumference or fat mass) and Africa. These terms and phrases were used in different combinations to be identified in titles and abstracts. Full-text articles were obtained and reviewed to identify those which met selection criteria below and data were extracted from relevant publications into tables appropriately.

Selection criteria

Studies which met the following criteria were considered relevant for inclusion:

• ∙ studies conducted in African countries;

• ∙ any study design;

• ∙ For observational studies

  • - Studies that described MNS (defined by reported dietary intakes, anthropometric data and biochemical indicators) in pregnant women of any age.

  • - Studies that associated MNS in pregnant women of any age with any fetal, birth, neonatal or infant outcome.

• ∙ For intervention studies

  • - Nutritional interventions done in pregnancy with dietary values and/or where biochemical indicators and fetal, birth, neonatal and/or infant outcome data from both the intervention and control group could be extracted.

  • - Nutritional interventions done in infancy with later adolescent or adult outcomes reported.

• ∙ Studies reporting data in a format that enabled daily mean or median nutrient intake for the population to be extracted.

Studies were excluded from the review according to the following criteria:

• ∙ studies conducted in animals;

• ∙ studies in subjects with health conditions that may have influenced dietary intake (i.e. gestational diabetes, coeliac disease);

• ∙ interventions (including any supplements) in which MNS and fetal, birth, neonatal and/or infancy outcomes were reported for the intervention group only.

MNS of pregnant African women (Table 1)

In all, 19 studies met the inclusion criteria.Reference Abebe, Bogale and Hambidge ^{18 –} Reference Stephens, Ofori, Quakyi, Wilson and Akanmori ³⁶ The publication year ranged from 2002 to 2014. The number of pregnant women examined in the studies ranged from 30 to 191,834 and the gestational age at MNS assessment varied between 18 and 39 weeks. Six studies used anthropometric measurements to describe MNS in pregnant women,Reference Assefa, Berhane and Worku ^{19 ,} Reference Elshibly and Schmalisch ^{23 –} Reference Hartikainen, Maleta, Kulmala and Ashorn ^{25 ,} Reference Liu, Joseph and Nkole ^{31 ,} Reference Nieuwoudt, van der Merwe, Harvey and Hall ³³ one used a biomarker of anaemia (haemoglobin (Hb)),Reference Stephens, Ofori, Quakyi, Wilson and Akanmori ³⁶ two used reported dietary intakesReference Belgnaoui and Belahsen ^{20 ,} Reference Huybregts, Roberfroid, Kolsteren and Van Camp ²⁶ and the remaining 10 used a combination of anthropometry, biomarkers and reported dietary intakesReference Abebe, Bogale and Hambidge ^{18 ,} Reference Changamire, Mwiru and Msamanga ^{21 ,} Reference Darwish, Mohamad, Gamal Al-Din, Elsayed and Ahmad ^{22 ,} Reference Jackson, Jackson and Yu ^{27 –} Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ^{30 ,} Reference Mostert, Steyn, Temple and Olwagen ^{32 ,} Reference Nti, Larweh and Gyemfua-Yeboah ^{34 ,} Reference Oguntona and Akinyele ³⁵ (Table 1).

Table 1 Results from observational studies describing maternal nutritional status (MNS) of pregnant African women

LBW, low birth weight; GWG, gestational weight gain; BMI, body mass index; MUAC, mid-upper arm circumference; Fe, iron; Ca, calcium; Zn, zinc; Vit, vitamin; Hb, haemoglobin; Ph, phosphorus; Mg, magnesium.

^a Based on serum Fe, ferritin, transferrin, Hb, haematocrit, mean corpuscular volume and red blood cell count.

^b Reported as kJ/day.

^c Median (interquartile range).

^d Median (25th, 75th percentiles).

^e Routine Fe supplementation received between 6 and 9 months.

Of the studies including anthropometric measures of MNS, five provided data for the mean/median body mass index (BMI). BMI varied from being within the normal range (18.5–24.9) in Tanzania,Reference Changamire, Mwiru and Msamanga ²¹ EthiopiaReference Abebe, Bogale and Hambidge ¹⁸ and Zambia,Reference Liu, Joseph and Nkole ³¹ to being within the overweight category (25.0–29.9) in South Africa,Reference Mostert, Steyn, Temple and Olwagen ³² SudanReference Elshibly and Schmalisch ^{23 ,} Reference Elshibly and Schmalisch ²⁴ and Zambia.Reference Liu, Joseph and Nkole ³¹ No studies reported mean/median BMI in either the underweight or obese categories. BMI was described according to WHO classification in two studies, with one describing a prevalence of 79.1% overweight and obesity in South African women and the other showing prevalence of 34.1 and 60.2% overweight and obesity in women who gave birth to normal weight and macrosomic babies, respectively, in Zambia.Reference Kesa and Oldewage-Theron ^{29 ,} Reference Liu, Joseph and Nkole ³¹ Weight gain was 228 g/week from ∼23 weeks gestational age in MalawiReference Hartikainen, Maleta, Kulmala and Ashorn ²⁵ and 1.06 kg/week during the third trimester in Liberia.Reference Jackson, Jackson and Yu ²⁷ In Sudan, the mean mid-upper arm circumference (MUAC) of pregnant women at delivery was 26.9 cm,Reference Elshibly and Schmalisch ^{23 ,} Reference Elshibly and Schmalisch ²⁴ whereas in Ethiopia 52.7% of women had an MUAC of <23 cmReference Assefa, Berhane and Worku ¹⁹ (Table 1).

Hb was used as a biomarker of iron status in pregnant women in five studies.Reference Changamire, Mwiru and Msamanga ²¹ All studies described mean Hb values above the threshold for diagnosis of anaemia in pregnant women (<11 g/dl), with the exception of one study in Tanzania.Reference Changamire, Mwiru and Msamanga ²¹ Based on this cut-off point 66.7 of rural and 26.7% of urban women were classified as anaemic in one study in Ghana and 32% were classified as anaemic in another study in the same setting.Reference Nti, Larweh and Gyemfua-Yeboah ^{34 ,} Reference Stephens, Ofori, Quakyi, Wilson and Akanmori ³⁶ Kenyan pregnant women had a 32% anaemia prevalence in one study,Reference Kamau-Mbuthia and Elmadfa ²⁸ whereas 42.2 and 21.8% of women from pastoral and farming communities, respectively, were diagnosed with anaemia in another study.Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ³⁰ In addition to anaemia diagnosed via Hb concentrations, Keverenge-Ettyang et al.Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ³⁰ also assessed iron stores in pregnant women using serum ferritin concentrations. Pregnant women from pastoral communities had significantly higher serum ferritin concentrations than those from farming communities, although the difference was relatively small (25.8 v. 24.4 µg/l, P<0.05). The prevalence of low maternal iron stores (serum ferritin<32 µg/l) was high in both groups (77% in pastoral and 85.9% in farming communities).Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ³⁰ A total of 27.9 and 24.2% of women from pastoral and farming communities, respectively, had low vitamin A status (serum retinol). Iron deficiency and iron deficiency anaemia (IDA) prevalence were 41.6 and 50%, respectively, in pregnant South African women based on a combination of biochemical markers (serum iron, ferritin, transferrin, Hb, haematocrit, mean corpuscular volume and red blood cell count).Reference Kesa and Oldewage-Theron ²⁹

Red cell folate concentrations were between 166 and 183 nmol/l in rural and between 158 and 177 nmol/l in urban Nigerian women.Reference Oguntona and Akinyele ³⁵ Mean calcium concentrations were 8.9 mg/dl in Egyptian pregnant women.Reference Darwish, Mohamad, Gamal Al-Din, Elsayed and Ahmad ²²

Of the 10 studies reporting dietary intake in pregnant women, most used 24 h recall and/or food frequency questionnaires as the assessment method,Reference Belgnaoui and Belahsen ^{20 ,} Reference Changamire, Mwiru and Msamanga ^{21 ,} Reference Huybregts, Roberfroid, Kolsteren and Van Camp ^{26 ,} Reference Kamau-Mbuthia and Elmadfa ^{28 ,} Reference Kesa and Oldewage-Theron ^{29 ,} Reference Mostert, Steyn, Temple and Olwagen ^{32 ,} Reference Nti, Larweh and Gyemfua-Yeboah ³⁴ whereas two used weighed food recordsReference Abebe, Bogale and Hambidge ^{18 ,} Reference Oguntona and Akinyele ³⁵ and one used a food survey questionnaire for calcium intake specifically.Reference Darwish, Mohamad, Gamal Al-Din, Elsayed and Ahmad ²² Mean energy intake ranged between 952 and 3981 kcal/day across study sites. Mean macronutrient intakes ranged as follows: carbohydrate 231–350 g/day, protein 15–104 g/day and fat 7–62 g/day; with the lowest intakes of all macronutrients found in the same Ethiopia population.Reference Abebe, Bogale and Hambidge ¹⁸ Mean intake of the key pregnancy micronutrients analysed ranged between 7–41 mg/day of iron, 194–424 µg/day of folate, 355–974 mg/day of calcium and 5–13 mg/day of zinc.

Associations between MNS (anthropometry) and fetal growth and birth, neonatal and infant outcomes (Table 2)

Eight studies met the inclusion criteria.Reference Assefa, Berhane and Worku ^{19 ,} Reference Elshibly and Schmalisch ^{23 –} Reference Hartikainen, Maleta, Kulmala and Ashorn ^{25 ,} Reference Jackson, Jackson and Yu ^{27 ,} Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ^{30 ,} Reference Liu, Joseph and Nkole ^{31 ,} Reference Nieuwoudt, van der Merwe, Harvey and Hall ³³ The publication year ranged from 2005 to 2014. Four studies followed a prospective cohort design,Reference Assefa, Berhane and Worku ^{19 ,} Reference Jackson, Jackson and Yu ^{27 ,} Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ^{30 ,} Reference Nieuwoudt, van der Merwe, Harvey and Hall ³³ two studies used retrospective dataReference Hartikainen, Maleta, Kulmala and Ashorn ^{25 ,} Reference Liu, Joseph and Nkole ³¹ and two were cross-sectional studies.Reference Elshibly and Schmalisch ^{23 ,} Reference Elshibly and Schmalisch ²⁴ The number of pregnant women included in the studies ranged from 80 to 191,834. Anthropometric measurements used to describe MNS in study participants included weight and height,Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ^{30 ,} Reference Nieuwoudt, van der Merwe, Harvey and Hall ³³ BMI,Reference Liu, Joseph and Nkole ³¹ gestational weight gain (GWG),Reference Hartikainen, Maleta, Kulmala and Ashorn ^{25 ,} Reference Jackson, Jackson and Yu ²⁷ MUACReference Assefa, Berhane and Worku ¹⁹ and lean body mass; with two studies describing all of these variables in pregnant women.Reference Elshibly and Schmalisch ^{23 ,} Reference Elshibly and Schmalisch ²⁴ All offspring outcomes were assessed at birth or within the first 30 days of life; with the most commonly examined outcomes being birth weight and gestational age (Table 2).

Table 2 Results from observational studies of the associations between maternal nutritional status (anthropometry) and fetal growth and birth, neonatal and infant outcomes

LBW, low birth weight; GWG, gestational weight gain; BMI, body mass index; MUAC, mid-upper arm circumference; MO, morbidly obese (BMI 40–49.9 kg/m²); SO, super obese (BMI⩾50 kg/m²); IUGR, intrauterine growth restriction; RR, risk ratio.

^a Median (interquartile range).

^b Median (25th, 75th percentiles.).

In Sudan, postpartum maternal weight and BMI were positively associated with birth weight (P<0.001), but neither variable predicted risk of LBW (<2500 g).Reference Elshibly and Schmalisch ²⁴ Although risk of LBW in Kenya was 2.4 times greater in infants born in farming than in pastoral communities, there was no difference in mean weight of the pregnant women during the third trimester of pregnancy.Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ³⁰ GWG in the second and third trimesters showed a strong seasonality effect in rural Malawian women, with those delivering in the rainy season gaining significantly less weight than those delivering in temperate/dry months (100–200 g/week compared with 250–300 g/week, P<0.001); however, this was not reflected as strongly in birth weight (P<0.05) and GWG was therefore only weakly correlated with birth weight (Pearson’s correlation coefficient 0.13; significance not reported).Reference Hartikainen, Maleta, Kulmala and Ashorn ²⁵ In contrast, although birth weight was correlated with maternal weight at 6 (r=0.54, P=0.01) and 9 months (r=0.53, P=0.01) in Liberia, there was a stronger, positive correlation with net weight gain between the two time points (β=0.059, P<0.001).Reference Jackson, Jackson and Yu ²⁷

In Zambia, where maternal overweight was more prevalent than the aforementioned studies, overweight and obesity were associated with 1.72 and 2.88 times greater odds of giving birth to a macrosomic infant, respectively.Reference Liu, Joseph and Nkole ³¹ In South African women with BMIs⩾40 kg/m², incidence of IUGR was significantly higher in those who had BMIs⩾50 kg/m² than those with BMIs between 40 and 49.9 kg/m².Reference Nieuwoudt, van der Merwe, Harvey and Hall ³³

MUAC was associated with birth weight in two studies, with an MUAC<23 cm (suggestive of maternal underweight)Reference Mohanty, Prasad and Srikanth Reddy ³⁷ increasing odds of LBW by 1.6 times.Reference Assefa, Berhane and Worku ^{19 ,} Reference Elshibly and Schmalisch ²³ In addition to the aforementioned findings, maternal height was identified as the strongest anthropometric predictor of neonatal outcomes in Liberia. Maternal height had positive associations with gestational age (P<0.002), limb length (P<0.001) and birth weight (P<0.001), whereas height<156 cm increased the relative risk of LBW by 52%. In the same study sample, positive associations were found between maternal lean body mass and birth weight, body length and body circumference within 24 h of birth (P<0.001).Reference Elshibly and Schmalisch ²³

The associations between MNS (nutritional biomarkers) and fetal growth and birth, neonatal and infant outcomes (Table 3)

Three studies met the inclusion criteria.Reference Jackson, Jackson and Yu ^{27 ,} Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ^{30 ,} Reference Stephens, Ofori, Quakyi, Wilson and Akanmori ³⁶ The publication years were 2006, 2010 and 2014. All studies included were prospective cohort studies and the sample size ranged between 80 and 320. All studies used biomarkers to assess anaemia and/or iron status of pregnant women, with two studies using Hb concentrations only and one study including haematocrit and serum ferritin concentrations.Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ³⁰ The latter also assessed maternal vitamin A status using serum retinol concentrations. Birth weight was the outcome of interest in all studies (Table 3).

Table 3 Results from observational studies of the associations between maternal nutritional status (biomarkers) and fetal growth and birth, neonatal and infant outcomes

LBW, low birth weight; Vit, vitamin; Hb, haemoglobin; Fe, iron.

^a Routine Fe supplementation received between 6 and 9 months.

Data from Kenya reports pregnant women from pastoral communities having lower Hb concentrations and higher anaemia prevalences than those from farming communities [(119 v. 124 g/l, P<0.05); (42.2 v. 21.8%, P<0.01)], respectively, but serum ferritin concentrations were higher in the pastoral community (25.8 v. 24.4 µg/l, P<0.05). Mean infant birth weight was significantly lower (2.9 v. 2.8 kg, P<0.01) and prevalence of LBW significantly higher (31.3 v. 16.8%, P<0.05) for babies born to mothers from farming than from pastoral communities.Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ³⁰ A weak correlation between maternal Hb concentrations at 6 months gestation and infant birth weight (P=0.042) was shown in Liberia; however, this was not significant at 9 months (all women had been routinely supplemented with 180 mg iron/day between the 6- and 9-month assessment).Reference Jackson, Jackson and Yu ²⁷ In Ghana, none of the women who were anaemic in the first trimester of pregnancy gave birth to LBW babies.Reference Stephens, Ofori, Quakyi, Wilson and Akanmori ³⁶

The associations between MNS (dietary intake) and fetal growth and birth, neonatal and infant outcomes

No studies were identified which met the inclusion criteria.

Randomized/quasi-randomized clinical trials on the associations between maternal nutritional interventions and fetal growth, birth, neonatal and infant outcomes (Table 4)

Six studies met the inclusion criteria.Reference Ceesay, Prentice and Cole ^{38 –} Reference Preziosi, Prual, Galan, Daouda, Boureima and Hercberg ⁴³ The publication years ranged between 1997 and 2011. Four studies were double-blind randomized controlled trials (RCTs),Reference Hawkesworth, Sawo and Fulford ^{40 –} Reference Preziosi, Prual, Galan, Daouda, Boureima and Hercberg ⁴³ one was a cluster RCTReference Ceesay, Prentice and Cole ³⁸ and one study used data from both a double-blind RCT and a cluster RCT.Reference Hawkesworth, Walker and Sawo ³⁹ The sample sizes ranged between 125 and 2100 and gestational age of the subjects at baseline ranged between 20 and 28 weeks. Interventions included iron,Reference Preziosi, Prual, Galan, Daouda, Boureima and Hercberg ⁴³ multiple micronutrient,Reference Kaestel, Michaelsen, Aaby and Friis ⁴² calciumReference Hawkesworth, Walker and Sawo ^{39 –} Reference Jarjou, Prentice and Sawo ⁴¹ and protein-energy supplementationReference Ceesay, Prentice and Cole ^{38 ,} Reference Hawkesworth, Walker and Sawo ³⁹ (Table 4).

Table 4 Results from randomized/quasi-randomized clinical trials on the associations between maternal nutritional interventions and fetal growth and birth, neonatal and infant outcomes

MNS, maternal nutritional status; LBW, low birth weight; Fe, iron; GWG, gestational weight gain; BMI, body mass index; OR, odds ratio.

Protein-energy supplementation in chronically undernourished Gambian women from 20 weeks gestational age was associated with 136 g higher pregnancy weight gain than in the control group (P<0.001).Reference Ceesay, Prentice and Cole ³⁸ Increases were higher in the hungry (201 g, P<0.001) than in the harvest season (94 g, P<0.01). Odds of perinatal mortality (death within the first 7 days of life) (OR: 0.54, 95% CI: 0.35; 0.85, P<0.01) and LBW (OR: 0.61, 95% CI: 0.47; 0.79, P<0.001) were, respectively, lower in the supplementation group. There was also a 3.1 mm increase in head circumference (P<0.01) in those who received the intervention.Reference Ceesay, Prentice and Cole ³⁸ During the follow-up study in 11–17-year olds, no differences in BMI, fat mass, lean mass, blood pressure, insulin or cholesterol concentrations were found between those whose mothers had received protein-energy supplementation during pregnancy and controls; however, those born to supplemented mothers had 0.05 mmol/l (95% CI: −0.10; −0.001 mmol/l) lower fasting glucose concentrations.Reference Hawkesworth, Walker and Sawo ³⁹

In Gambian women, no differences in weight, body length, head circumference or bone mineral content between infants born to women who received calcium supplementation and those who received a placebo from 20 weeks gestational age were observed. Follow-up of infants at 5–10 years of age showed no differences in blood pressure and no interaction between BMI and calcium supplementation for blood pressure variables.Reference Hawkesworth, Walker and Sawo ^{39 ,} Reference Hawkesworth, Sawo and Fulford ⁴⁰

A trial of iron supplementation to a cohort of pregnant women with a high anaemia prevalence in Niger found no differences in birth weight between babies born to the intervention and control groups.Reference Preziosi, Prual, Galan, Daouda, Boureima and Hercberg ⁴³ Birth length and Apgar scores were significantly higher in babies born to supplemented mothers; however, the difference in length did not persist at 3 and 6 months. Serum ferritin concentrations at 3 and 6 months of age were higher in infants whose mothers received iron supplementation compared with mothers who received the placebo (P<0.05).Reference Preziosi, Prual, Galan, Daouda, Boureima and Hercberg ⁴³

Pregnant women in Guinea-Bissau received either one of two possible interventions: a tablet with one recommended daily allowance (RDA) of 15 micronutrients (MN-1) or a tablet with two RDAs of the same micronutrients (but one RDA of iron) (MN-2), or a standard iron-folic acid supplement (control group) from ∼22 weeks gestation.Reference Kaestel, Michaelsen, Aaby and Friis ⁴² Mean birth weight was 53 and 95 g higher in the MN-1 and MN-2 groups, respectively, than in the control group, suggesting a dose–response effect of supplementation. Supplementation had a positive effect on LBW, with 10.1, 12 and 13.6% LBW prevalence found in the MN-2, MN-1 and control groups, respectively; however, this was not significant (P=0.33). Birth weight was 218 g higher and risk of LBW 69% lower for babies born to anaemic women in the MN-2 group compared with the control group.Reference Kaestel, Michaelsen, Aaby and Friis ⁴²

Randomized/quasi-randomized clinical trials on the associations between nutritional interventions in the first 2 years of life and any adolescent and/or adult health outcomes

No studies were identified which met the inclusion criteria.

MNS in Africa

Using BMI most African women in the reported studies were within the normal weight to overweight category during pregnancy, with maternal overweight or obesity being more prevalent than underweight. This was supported by high energy intakes in some countries; however, mean energy intakes varied greatly between populations. Carbohydrate, protein and fat contributed between 58–87, 5.9–14.5 and 6.3–27.9%, respectively, to total energy intake across study sites. The most prevalent micronutrient deficiency in African pregnant women was iron.

Studies included in this review indicate a low burden of maternal underweight, and comparatively high overweight and obesity prevalence, typical of the epidemiological health transition across African countries. Although GWG was approximately half the Institute of Medicine (IOM) recommended level for normal weight women in Malawi,Reference Hartikainen, Maleta, Kulmala and Ashorn ²⁵ much higher weekly gain than recommended for any BMI category was found in Liberian pregnant women.Reference Jackson, Jackson and Yu ^{27 , 44} Though the above findings provide good proxies for maternal obesity status, interpretation of the findings should be done with care, as a limitation exits in that obesity status (being underweight, normal, overweight or obese) was categorized by using BMI cutoffs of non-pregnant women.

Nutritional biomarkers showed a persisting high prevalence of micronutrient deficiencies in pregnant African women. Data suggest anaemia and/or iron deficiency prevalence to be high. Although comparison between rural and urban sites in Ghana suggests significantly higher anaemia prevalence in rural women, close to 30% of pregnant women were anaemic in urban settings.Reference Nti, Larweh and Gyemfua-Yeboah ³⁴ This is much higher than the prevalence seen in HICs (Europe: 16.2%, America and the Caribbean: 15.2%) and other LMIC (Asia: 19.8%) settings.Reference Black, Victora and Walker ⁴⁵ This may be owing to a chronic intake of low absorbable iron and insufficient iron stores to support both maternal and fetal requirements or to high levels of infection in African communities; or a combination of both.Reference Abu-Saad and Fraser ⁶

Reported dietary intakes of pregnant women varied significantly across African countries, with studies showing energy consumption below and above the American Dietetic Association recommended range of 2200–2900 kcal/day.Reference Kaiser and Allen ⁴⁶ However, the shift towards higher energy intakes in populations with traditionally low food access was not reflected in adequate protein intake, which was lower than the IOM’s RDA (71 g) in all but one study. ⁴⁷ Mean dietary iron intakes were much lower than the IOM RDA of 27 mg/day ⁴⁸ in all but two studies.Reference Abebe, Bogale and Hambidge ^{18 ,} Reference Huybregts, Roberfroid, Kolsteren and Van Camp ²⁶ Folate intake was much lower, on average, than the 600 µg/day recommended for pregnant women,Reference Bailey ⁴⁹ with most studies reporting intakes less than half of the recommended intake. Majority of the women studied either did not receive or did not comply with micronutrient supplementation during pregnancy, even in countries where iron and/or folic acid supplementation should have been a routine part of antenatal care. Higher energy consumption, coupled with inadequate protein and micronutrient intakes, may be a result of poor diet quality and/or food availability in communities in transition. This was demonstrated in South Africa where one study showed pregnant women to consume predominantly cereal-based diets high in energy and refined sugar, with low intakes of more expensive protein/micronutrient-rich foods such as meat, poultry and seafood, as well as legumes and non-starchy vegetables.Reference Kesa and Oldewage-Theron ²⁹

Associations of MNS (using anthropometric parameters) and fetal growth and birth, neonatal and infant outcomes

Maternal weight, BMI and weight gain during pregnancy were positively associated with birth weight in African studies. However, maternal overweight and obesity increased the risk of macrosomia in ZambiaReference Liu, Joseph and Nkole ³¹ and higher BMIs were associated with increased risk of IUGR in a sample of very obese women from South Africa (BMI>40 kg/m²).Reference Nieuwoudt, van der Merwe, Harvey and Hall ³³

These findings are consistent with studies from other parts of the world. A systematic review including data from both HICs and LMICs showed a significant risk of LBW in women who were underweight during pregnancy compared with those who were within normal weight categoriesReference Han, Mulla, Beyene, Liao and McDonald ⁹ and GWG has been positively associated with birth weight in a number of studies.Reference Frederick, Williams, Sales, Martin and Killien ^{50 –} Reference Siega-Riz, Viswanathan and Moos ⁵² Substantial evidence supports the association between maternal obesity and macrosomia, with a two to three-fold increase in risk of macrosomia being observed in obese women.Reference Kerrigan and Kingdon ^{2 ,} Reference Leddy, Power and Schulkin ^{3 ,} Reference Ruager-Martin, Hyde and Modi ¹⁰ There is also evidence to support the association between maternal obesity and IUGR; however, fewer studies have documented this.Reference Radulescu, Munteanu, Popa and Cirstoiu ^{53 ,} Reference Rajasingam, Seed, Briley, Shennan and Poston ⁵⁴

Although the findings of this review have been supported by literature from other settings, the strength of and comparability between the included studies is limited owing to differences in study design, exposure variables and sample sizes that were relatively low in prospective cohort designs. In addition, the timing in assessment of anthropometric parameters in pregnancy and outcome measurements varied greatly between studies, with maternal assessments being done between the first antenatal visit and delivery across studies and birth outcome measurements being taken any time between birth and the first 30 days of life.

Although the underweight prevalence was low overall in African settings, risk of adverse fetal and birth outcomes remained high in populations where low pre-pregnancy weight is a key issue, for example, in Ethiopia where 52.7% of women had MUAC measurements <23 cm, LBW prevalence was high (28.3%). However, as maternal overweight and obesity continue to rise, high pre-pregnancy weight and excessive GWG and their associated risks should become the pivotal focus for maternal and child nutrition.

Associations of MNS (nutritional biomarkers) and fetal growth and birth, neonatal and infant outcomes

Data on the associations between nutritional biomarkers and outcomes of interest were very limited. The limited data available seem to suggest associations between low Hb and serum ferritin concentrations with lower birth weights in African settings.Reference Jackson, Jackson and Yu ^{27 ,} Reference Keverenge-Ettyang, van Marken Lichtenbelt, Esamai and Saris ³⁰

Global evidence associating nutritional biomarkers with outcomes of interest show mixed results. Maternal anaemia in Indians was associated with increased risk of LBW and IDA predicted a three times higher risk of preterm birth.Reference Finkelstein, Duggan and Thomas ⁵⁵ Low maternal Hb, but not serum ferritin concentrations, were associated with lower birth weight in Iran.Reference Samimi, Asemi and Taghizadeh ⁵⁶ However, in a multicentre study across four HICs (New Zealand, Australia, England and Ireland), as well as in Sri Lanka (LMIC), there were no associations observed between anaemia (Hb<11 g/dl) and risk of preterm birth, LBW or small for gestational age (SGA) infants.Reference Abeysena, Jayawardana and Seneviratne ^{57 ,} Reference Masukume, Khashan, Kenny, Baker and Nelson ⁵⁸

Although all included studies focussed on the association between anaemia and birth weight using prospective cohort designs, the late assessment of biomarker status in two of the three studies provided poor proxies of pre-pregnancy status and the variation in timing of measurements between studies limited comparability. Sample sizes were low in all studies, which may have limited the power to detect associations between the biomarker(s) and outcome(s) of interest.

More evidence is needed to understand the associations between maternal micronutrient status and deficiencies on outcomes of interest in the first 1000 days; however, use of individual biomarkers of nutritional status in isolation may be impractical in Africa where diet quality is poor and pregnant women are likely to experience multiple nutrient deficiencies. Thus, identifying nutritional biomarker patterns using dimension reduction techniques could be essential to employ in such studies.

Associations of MNS (reported dietary intakes) and fetal growth and birth, neonatal and infant outcomes

Although no articles were retrieved for Africa on associations between reported dietary intakes and outcomes of interest, the use of reported dietary assessment (DA) has a number of challenges. Repeated 24 h recalls and food frequency questionnaires are the most commonly used methods for assessing habitual dietary intakes in Africa.Reference Pisa, Landais and Margetts ⁵⁹ The inherent errors associated with reported dietary intakes and the strengths and limitations of different DA methods cannot be ignored.Reference Pisa, Landais and Margetts ⁵⁹ Limitations include recall bias, assuming temporarily regular eating habits, seasonality and providing inaccurate estimations of portion size, etc.Reference Johnson, Soultanakis and Matthews ^{60 –} Reference Wrieden, Peace, Armstrong and Barton ⁶² However, very few DA tools used in Africa have been validated or tested for reliability that presents a huge challenge for effective assessment and monitoring of dietary intake, as well as for comparison of intakes within and between African settings.Reference Pisa, Landais and Margetts ⁵⁹

Randomized clinical trials of maternal nutritional interventions and fetal growth and birth, neonatal and infant outcomes

Evidence from an African study suggests positive associations between protein-energy supplementation during pregnancy and higher GWG, birth weight and lower risk of perinatal mortality.Reference Ceesay, Prentice and Cole ³⁸ However, no long-term effects were seen on CVD risk during the 11–17-year follow-up.Reference Hawkesworth, Walker and Sawo ³⁹ Multiple micronutrient supplementations had a dose–response effect on birth weight and significantly reduced LBW risk in anaemic women.Reference Kaestel, Michaelsen, Aaby and Friis ⁴² Although iron supplementation was associated with an increase in birth length, no improvement in birth weight was found.Reference Preziosi, Prual, Galan, Daouda, Boureima and Hercberg ⁴³ Prenatal calcium supplementation had no effect on any birth, neonatal, infant or childhood health outcomes.Reference Hawkesworth, Walker and Sawo ^{39 –} Reference Jarjou, Prentice and Sawo ⁴¹

A review on protein-energy supplementation trials including both HICs and LMICs showed positive effects on birth weight in the supplemented compared with control groups, with the greatest effects seen in undernourished populations.Reference Imdad and Bhutta ⁶³ This supports the findings of the Gambian study where pregnant women were chronically undernourished at baseline.Reference Ceesay, Prentice and Cole ³⁸ Similarly, a meta-analysis on multi-micronutrient supplementation trialsReference Ramakrishnan, Grant and Goldenberg ⁶⁴ supported the data from Guinea-BissauReference Kaestel, Michaelsen, Aaby and Friis ⁴² by showing significant reductions in LBW, SGA incidence and increased mean birth weight for women in the intervention compared with the control group (mostly receiving iron-folate supplements).Reference Ramakrishnan, Grant and Goldenberg ⁶⁴ Data from the iron supplementation trial in NigerReference Preziosi, Prual, Galan, Daouda, Boureima and Hercberg ⁴³ contradicted the general findings from a meta-analysis which showed that daily iron supplementation during pregnancy (alone or in combination with folate) reduced incidence of LBW by 20% compared with controls.Reference Imdad and Bhutta ⁶⁵ The effects of supplementation were most pronounced in populations with higher baseline anaemia prevalence.Reference Imdad and Bhutta ⁶⁵ Although positive effects on birth length were seen in the supplementation group in Niger, the lack of improvement in birth weight in this population with high anaemia prevalence is not a common finding compared with the literature. However, this finding could be, in part, attributed to the late start of the intervention during pregnancy (38 weeks, ±21 days) and/or the small sample size.Reference Preziosi, Prual, Galan, Daouda, Boureima and Hercberg ⁴³ No studies in Africa were found that suggested any long-term benefits of nutritional supplementation during pregnancy.

RCTs are considered the most robust designs for assessing the relationship between exposure and outcome, because they ensure comparability between those exposed and those unexposed to the intervention and allow for causal links to be made as the intervention always precedes the outcome of interest. However, variability between RCT designs can alter the strength of individual studies. The following are important factors of concern associated with maternal nutritional intervention studies that make comparability of findings difficult: (i) sample size, (ii) dose of intervention, (iii) timing of intervention during pregnancy, (iv) baseline nutritional status of pregnant woman and (v) an appropriate control group. Cumulatively, the data available seem to suggest significant benefits of macronutrient and/or micronutrient supplementation during pregnancy on fetal/birth outcome (specifically birth weight), particularly in undernourished women.

The results presented in this review illustrate that data available for Africa ranges from 18 weeks gestational age onwards. This highlights that an important critical phase (<18 weeks) has not been investigated in this setting.

The most important nutrients of concern for the first 1000 days for African women

Energy and nutrient requirements increase during pregnancy in order to meet the needs of both the mother and the growing fetus. Inadequate intakes of macro- and/or micronutrients before and during pregnancy result in limited growth and development and therefore poor pregnancy outcomes. Nutrient sufficiency is similarly required during early infancy to prevent growth faltering. Certain nutrients are of particular importance, owing to the critical functions that they perform and the plasticity during the first 1000 days of life. Energy requirements increase during pregnancy to support increases in basal metabolic rate as a result of growth and expansion of new and existing tissue (fetus, placenta and maternal tissues), as well as the higher work rate of the maternal cardiovascular, respiratory and renal systems. Adequate energy is also needed to support periods of rapid growth and development in the first 2 years of life. Protein requirements are high during pregnancy and infancy for deposition and maintenance of maternal and fetal tissue.Reference Abu-Saad and Fraser ^{6 ,} Reference Dupont ⁶⁶ Omega-3 and omega-6 fatty acids are essential to new tissue formation, owing to their structural role in cell membranes, with omega-3 fatty acids being particularly important for brain and central nervous system development.Reference Greenberg, Bell and Ausdal ⁶⁷ Micronutrients of key concern in pregnancy are iron and folate, as they are unlikely to be in sufficient supply from the diet. Additional iron is required to support the increase in red cell mass and ensure sufficient oxygen supply during tissue synthesis and growth. Folate is an important co-factor in cellular function, including DNA and nucleic acid synthesis and cell division.Reference Abu-Saad and Fraser ^{6 ,} Reference Scholl ^{68 ,} Reference Scholl and Johnson ⁶⁹

Known consequences associated with poor MNS

Inadequate maternal nutrition – underweight and overweight – as well as micronutrient insufficiency has been strongly linked with adverse maternal and infant outcomes, with both short- and long-term consequences. Maternal obesity and adiposity and high GWG are associated with increased risk of GDM, pre-eclampsia, maternal weight retention postpartum and poor infant outcomes such as prolonged labour, birth trauma, neonatal death and contrasting burdens of both macrosomia and SGA. Higher neonatal fat mass has been associated with adiposity in childhood and adulthood and therefore increased metabolic risk in later life.Reference Ruager-Martin, Hyde and Modi ^{10 ,} Reference O’Reilly and Reynolds ¹⁵ Premature delivery has been shown to be associated with both maternal underweight and overweight and is strongly associated with increased risk of perinatal morbidity and mortality, as well as impaired cognitive and emotional development later in childhood and adolescence.Reference Cnattingius, Villamor and Johansson ^{8 ,} Reference Han, Mulla, Beyene, Liao and McDonald ^{9 ,} Reference Saigal and Doyle ⁷⁰ Low maternal weight-for-height, poor GWG and micronutrient deficiencies such as IDA increase risk of IUGR, which is associated with neonatal mortality in the short term and sub-optimal growth and development in the long term; for example, in cognition, learning disabilities, academic achievement and psychosocial maturation.Reference Ergaz, Avgil and Ornoy ⁷¹ For those infants defined as SGA, neonatal mortality risk is higher than those born appropriate for gestational age, even if born at term.Reference Abu-Saad and Fraser ⁶ LBW, a result of preterm birth and/or growth restriction in utero, is associated with increased risk of perinatal morbidity and mortality, as well as of long-term health risk. Sufficient evidence exists to suggest that impaired growth in utero increases long-term risk of NCDs such as T2DM, hypertension and CVD, with the highest level of risk being seen in those who subsequently experience rapid and/or excessive of weight gain.Reference Abu-Saad and Fraser ⁶

Challenges of appropriate interventions in the first 1000 days to reduce childhood obesity and adult NCDs in African women

MNS, childhood obesity and adult NCD risk is complex and influenced by multiple factors at various life stages, making it difficult to reverse once highly prevalent in populations. Exposure to a poor nutrition environment in the first 1000 days (critical periods of plasticity) seems to have significant effects on body function, metabolism and a programming phenotypic effect, thereby influencing susceptibility to obesity, as well as to NCDs, in the longer term. This is of critical importance in the African setting where maternal obesity, coupled with poor micronutrient status and diet quality, continues to grow.Reference Black, Victora and Walker ^{45 ,} Reference Vorster, Kruger and Margetts ^{72 ,} Reference Mokhtar, Elati and Chabir ⁷³ Although the plastic nature of this period makes it vulnerable to poor environmental exposures, it also provides a unique window for intervention. Ensuring optimal growth and development during this window, when women are highly motivated and tend to experience greater contact with health services, should therefore be prioritized in Africa to improve long-term health trajectories.Reference Gillman and Ludwig ⁷⁴ The main challenges for appropriate nutritional interventions in first 1000 days include (i) when to intervene to get the best returns (pre-pregnancy v. early pregnancy v. after birth v. infancy) and (ii) which nutrients and what doses to include. The Lancet series on maternal and child nutrition has provided a new conceptual framework that shows and elucidates on the means to optimum fetal and child growth and development.Reference Black, Victora and Walker ⁴⁵ This framework outlines the dietary, behavioural and health determinants of optimum nutrition, growth and development, and how they are affected by various underlying conditions, which are in turn shaped by economic and social conditions, national and global contexts, capacity, resources and governance. In addition, the series outlines and discusses how determinants can be changed to enhance maternal and childhood outcomes, including nutrition-specific interventions that address the immediate and underlying causes of malnutrition.Reference Bhutta, Das and Rizvi ^{75 ,} Reference Ruel and Alderman ⁷⁶

There is a lack of data associating MNS with outcomes beyond birth in Africa. A need exists for longitudinal data from pregnancy through infancy to 2 years of age, and beyond. Without this evidence we cannot adequately influence policy or strengthen health systems.