Effects of diet and exercise interventions compared with no diet and exercise interventions (EB Table A.1.1)

The evidence on the effects of healthy eating and exercise interventions was derived from a Cochrane review that included 65 randomized controlled trials (RCTs), mostly conducted in HICs (44). Thirty-four trials recruited women from the general population (i.e. women of a wide range of BMIs at baseline), 24 trials recruited overweight and/or obese women and seven recruited women defined as being at high risk of gestational diabetes. In total, 49 RCTs involving 11 444 women contributed data to the review’s meta-analyses. Diet interventions were defined as a special selection of food or energy intake to which a participant was restricted, which were most commonly “healthy eating” types of diets. Exercise interventions were defined by reviewers as any activity requiring physical effort, carried out to sustain or improve health or fitness, and these were either prescribed/unsupervised (e.g. 30 minutes of daily walking), supervised (e.g. a weekly supervised group exercise class) or both. These interventions were usually compared with “standard ANC” and aimed to prevent excessive gestational weight gain (EGWG).

Most trials recruited women between 10 and 20 weeks of gestation. There was substantial variation in the number of contacts (i.e. counselling/exercise sessions), type of intervention and method of delivery. Data were grouped according to the type of intervention (i.e. diet only, exercise only, diet and exercise counselling, diet and supervised exercise) and the average effects across trials were estimated using the random effects model. Separate analyses were performed according to type of intervention and the risk of weight-related complications. Most data in the overall analyses were derived from trials of combined diet and exercise interventions.

Values

Please see “Women’s values” in section 3.A: Background (p. 15).

Resources

Cost implications of diet and exercise interventions for health services are highly variable. For example, supervised diet and exercise interventions can have high associated costs, mainly due to staff costs for time spent supervising, while counselling interventions might have relatively low costs. For pregnant women, the interventions might also have resource implications in terms of transport costs, time off work and child-minding costs, particularly if the intervention requires additional antenatal visits.

Equity

Most of the evidence came from trials conducted in HICs. Recent studies have reported a shift in the burden of overweight and obesity from advantaged to disadvantaged populations (32). Such a trend increases the risk of associated pregnancy complications, as well as cardiometabolic problems, among pregnant women from disadvantaged populations. These risks might be further exacerbated among women in low-resource community settings, as these settings may not be equipped to deal with complications.

Acceptability

Qualitative evidence indicates that women in a variety of settings tend to view ANC as a source of knowledge and information and that they generally appreciate any advice (including dietary or nutritional) that may lead to a healthy baby and a positive pregnancy experience (high confidence in the evidence) (22). It also suggests that women may be less likely to engage with health services if advice is delivered in a hurried or didactic manner (high confidence in the evidence) (22). Therefore, these types of interventions are more likely to be acceptable if the interventions are delivered in an unhurried and supportive way, which may also facilitate better engagement with ANC services. Qualitative evidence on health-care providers’ views of ANC suggests that they may be keen to offer general health-care advice and specific pregnancy-related information (low confidence in the evidence) but they sometimes feel they do not have the appropriate training and lack the resources and time to deliver the service in the informative, supportive and caring manner that women want (high confidence in the evidence) (45).

Feasibility

In a number of LMIC settings, providers feel that a lack of resources may limit implementation of recommended interventions (high confidence in the evidence) (45).

Effects of nutritional education to increase energy and protein intake versus no nutritional education intervention (EB Table A.1.2)

Evidence on the effects of nutritional education was derived from a Cochrane review (47). Five trials conducted between 1975 and 2013 in Bangladesh, Greece and the USA, involving 1090 pregnant women, contributed data to this comparison. Nutritional education interventions were delivered one-to-one or in group classes and included education to improve the “quality” of diet, increase energy and protein intake, or improve knowledge of the nutritional value of different foods, including energy, protein, vitamins and iron. The Bangladesh study also involved cookery demonstrations.

Values

Please see “Women’s values” in section 3.A: Background (p. 15).

Resources

Resource costs are variable and mainly related to staffing and counselling time.

Equity

In many LMICs, pregnancy outcomes and ANC coverage are worse among women who are poor, least educated and residing in rural areas (29). Many low-income countries still struggle with widespread poverty and hunger, particularly among rural populations (48). Findings from a study of antenatal food supplementation and micronutrient supplements in rural Bangladesh suggest that food supplementation interventions might be associated with better ANC adherence among women with less education but not among those with more education (49). Therefore, providing antenatal food supplements could help to address inequalities by improving maternal nutritional status and increasing ANC coverage among disadvantaged women.

Acceptability

Qualitative evidence indicates that women in a variety of settings tend to view ANC as a source of knowledge and information and that they generally appreciate any advice (including dietary or nutritional) that may lead to a healthy baby and a positive pregnancy experience (high confidence in the evidence) (22). It also suggests that women may be less likely to engage with health services if advice is delivered in a hurried or didactic manner (high confidence in the evidence) (22). Therefore, these types of interventions are more likely to be acceptable if the interventions are delivered in an unhurried and supportive way, which may also facilitate better engagement with ANC services. Qualitative evidence on health-care providers’ views of ANC suggests that they may be keen to offer general health-care advice and specific pregnancy-related information (low confidence in the evidence) but they sometimes feel they do not have the appropriate training and lack the resources and time to deliver the service in the informative, supportive and caring manner that women want (high confidence in the evidence) (45).

Feasibility

In a number of LMIC settings, providers feel that a lack of resources may limit implementation of recommended interventions (high confidence in the evidence) (45).

Effects of balanced energy and protein supplements compared with no supplements or placebo (EB Table A.1.3)

Evidence on the effects of balanced energy and protein supplements compared with no supplementation or placebo was derived from a Cochrane review (47). Twelve trials, involving 6705 women, were included in this comparison. Most data were derived from trials conducted in LMICs, including Burkina Faso, Colombia, Gambia, Ghana, India, Indonesia, South Africa and Taiwan, China. The balanced energy and protein supplements used were in various forms, including fortified beverages, biscuits and powders.

Values

Please see “Women’s values” in section 3.A: Background (p. 15).

Resources

The cost of balanced energy and protein supplements is relatively high. There may also be cost implications with respect to transport, storage and training.

Equity

In many LMICs, pregnancy outcomes and ANC coverage are worse among women who are poor, least educated and residing in rural areas (29). Many low-income countries still struggle with widespread poverty and hunger, particularly among rural populations (48). Findings from a study of antenatal food supplementation and micronutrient supplements in rural Bangladesh suggest that food supplementation interventions might be associated with better ANC adherence among women with less education but not among those with more education (49). Therefore, providing antenatal food supplements could help to address inequalities by improving maternal nutritional status and increasing ANC coverage among disadvantaged women.

Acceptability

Qualitative evidence indicates that women in a variety of settings tend to view ANC as a source of knowledge and information and that they generally appreciate any advice (including dietary or nutritional) that may lead to a healthy baby and a positive pregnancy experience (high confidence in the evidence) (22). It also suggests that women may be less likely to engage with health services if advice is delivered in a hurried or didactic manner (high confidence in the evidence) (22). Therefore, these types of interventions are more likely to be acceptable if the interventions are delivered in an unhurried and supportive way, which may also facilitate better engagement with ANC services. Qualitative evidence on health-care providers’ views of ANC suggests that they may be keen to offer general health-care advice and specific pregnancy-related information (low confidence in the evidence) but they sometimes feel they do not have the appropriate training and lack the resources and time to deliver the service in the informative, supportive and caring manner that women want (high confidence in the evidence) (45).

Feasibility

Providing balanced protein and energy supplements may be associated with logistical issues, as supplements are bulky and will require adequate transport and storage facilities to ensure continual supplies. Qualitative evidence from LMIC settings indicates that providers feel that a lack of resources may limit implementation of recommended interventions (high confidence in the evidence) (45).

Effects of high-protein supplementation compared with controls (EB Table A.1.4)

Evidence on the effects of high-protein supplementation was derived from the same Cochrane review as for Recommendations A.1.2 and A.1.3 (47). The review included one trial of high-protein supplementation compared with a micronutrient supplement conducted in the 1970s, involving 1051 low-income, black women in the USA.

Values

Please see “Women’s values” in section 3.A: Background (p. 15).

Resources

The cost of high-protein supplements is relatively high. There may also be cost implications with respect to transport, storage and training.

Equity

In many LMICs, pregnancy outcomes and ANC coverage are worse among women who are poor, least educated and residing in rural areas (29). Many low-income countries still struggle with widespread poverty and hunger, particularly among rural populations (48). Therefore, providing antenatal food supplements could help to address inequalities by improving maternal nutritional status and increasing ANC coverage among disadvantaged women.

Acceptability

Qualitative evidence indicates that women in a variety of settings tend to view ANC as a source of knowledge and information and that they generally appreciate any advice (including dietary or nutritional) that may lead to a healthy baby and a positive pregnancy experience (high confidence in the evidence) (22). It also suggests that women may be less likely to engage with health services if advice is delivered in a hurried or didactic manner (high confidence in the evidence) (22). Qualitative evidence on health-care providers’ views of ANC suggests that they may be keen to offer general health-care advice and specific pregnancy-related information (low confidence in the evidence) but they sometimes feel they do not have the appropriate training and lack the resources and time to deliver the service in the informative, supportive and caring manner that women want (high confidence in the evidence) (45).

Feasibility

Providing high-protein supplements may be associated with logistical issues, as supplements are bulky and will require adequate transport and storage facilities to ensure continual supplies. Qualitative evidence from LMIC settings indicates that providers feel that a lack of resources may limit implementation of recommended interventions (high confidence in the evidence) (45).

Effects of any daily iron and folic acid supplements compared with no daily iron and folic acid supplements (EB Table A.2.1)

The evidence on the effects of daily iron and/or folic acid was derived from a Cochrane review of 61 trials conducted in low-, middle- and high-income countries (54). Twenty-three trials were conducted in countries with some malaria risk, of which two reported malaria outcomes. Overall, 44 trials involving 43 274 women contributed data to the review’s meta-analyses. The trials compared daily oral iron supplementation, with or without folic acid or other vitamin and mineral supplements, with various control groups (folic acid only, placebo, no intervention, other vitamin and mineral supplements without iron or folic acid). Most of the evidence was derived from studies comparing iron supplementation with no iron supplementation. In most trials, women began taking supplements before 20 weeks of gestation and continued taking supplements until delivery. The most commonly used dose of elemental iron was 60 mg daily (range: 30–240 mg) and that of folic acid was 400 μg daily.

Values

Please see “Women’s values” in section 3.A: Background (p. 15).

Resources

Daily iron and folic acid supplements are relatively low cost, at less than 1 United States dollar (US$ 1) per pregnant woman (27).

Equity

Iron deficiency and parasitic infections are more common in LMICs and disadvantaged populations. Poor, rural and least-educated populations also experience the highest maternal, infant and child mortality (29). Increasing coverage of effective nutritional interventions to prevent anaemia, particularly among disadvantaged populations, might help to address maternal and newborn health inequalities.

Acceptability

Qualitative evidence suggests that the availability of iron supplements may actively encourage women to engage with ANC providers (low confidence in the evidence) (22). However, where there are additional costs associated with supplementation or where the supplements may be unavailable (because of resource constraints) women are less likely to engage with ANC services (high confidence in the evidence). Lower doses of iron may be associated with fewer side-effects and therefore may be more acceptable to women than higher doses.

Feasibility

Qualitative evidence about the views of health-care providers suggests that resource constraints, both in terms of the availability of the supplements and the lack of suitably trained staff to deliver them, may limit implementation (high confidence in the evidence) (45).

Effects of intermittent iron and folic acid supplements compared with daily iron and folic acid supplements (EB Table A.2.2)

The evidence on the effects of intermittent iron and folic acid was derived from a Cochrane review that included 27 trials from 15 countries; however, only 21 trials (involving 5490 women) contributed data to the review’s meta-analyses (56). All trials were conducted in LMICs with some degree of malaria risk (Argentina, Bangladesh, China, Guatemala, India, Indonesia, the Islamic Republic of Iran, Malawi, Malaysia, Mexico, Pakistan, Republic of Korea, Sri Lanka, Thailand and Viet Nam); however, only one trial specifically reported that it was conducted in a malaria-endemic area.

Most of the intermittent iron regimens involved women taking weekly supplements, most commonly 120 mg elemental iron per week (range: 80–200 mg weekly), which was compared with daily regimens, most commonly 60 mg elemental iron daily (range: 40–120 mg daily). Where folic acid was also provided in the trials, it was administered weekly in the intermittent supplement groups (range: 400–3500 μg weekly) compared with the usual standard daily dose for control groups.

Values

Please see “Women’s values” in section 3.A: Background (p. 15).

Resources

Intermittent iron and folic acid supplementation might cost a little less than daily iron and folic acid supplementation due to the lower total weekly dose of iron.

Equity

Intermittent iron and folic acid supplementation may have less impact on health inequalities than daily iron and folic acid supplementation, as anaemia is more common in disadvantaged populations.

Acceptability

Qualitative evidence suggests that the availability of iron supplements may actively encourage women to engage with ANC providers (low confidence in the evidence) (22). However, where there are additional costs associated with supplementation or where the supplements may be unavailable (because of resource constraints) women are less likely to engage with ANC services (high confidence in the evidence). Women may find intermittent iron supplementation more acceptable than daily iron supplementation, particularly if they experience side-effects with daily iron supplements.

Feasibility

Intermittent iron may be more feasible in some low-resource settings if it costs less than daily iron.

Maternal outcomes

High-certainty evidence shows that calcium supplementation does not have important effects on maternal anaemia (1 trial, 1098 women; RR: 1.04, 95% CI: 0.90–1.22) or caesarean section rates (9 trials, 7440 women; RR: 0.99, 95% CI: 0.89–1.10). Moderate-certainty evidence indicates that calcium supplementation probably has little or no effect on maternal mortality (2 trials, 8974 women; RR: 0.29, 95% CI: 0.06–1.38) and probably makes little or no difference to the risk of urinary tract infections (3 trials, 1743 women; RR: 0.95, 95% CI: 0.69–1.30). Low-certainty evidence suggests that calcium supplementation may make little or no difference to maternal weight gain (3 trials; MD: −29.46 g per week, 95% CI: −119.80 to 60.89 g per week). Maternal satisfaction was not reported in any of the trials included in the Cochrane review.

Side-effects: Calcium supplementation makes little or no difference to the risk of “any side-effect”, a composite outcome including headache, vomiting, backache, swelling, vaginal and urinary complaints, dyspepsia and abdominal pain (1 trial, 8312 women; RR: 1.02, 95% CI: 0.93–1.12), and probably makes little or no difference to the risk of urinary stones (3 trials, 13 419 women; RR: 1.11, 95% CI: 0.48–2.54), renal colic (1 trial, 8312 women; RR: 1.67, 95% CI: 0.40–6.99) and impaired renal function (1 trial, 4589 women; RR: 0.91, 95% CI: 0.51–1.64), all assessed as moderate-certainty evidence. Low-certainty evidence suggests that it may have little or no effect on the risk of gallstones (1 trial, 518 women; RR: 1.35, 95% CI: 0.48–3.85).

Fetal and neonatal outcomes

Calcium supplementation probably has little or no effect on low-birth-weight babies (< 2500 g), as indicated by evidence that was of moderate certainty due to inconsistency (6 trials, 14 162 women; RR: 0.93, 95% CI: 0.81–1.07). Low-certainty evidence suggests that it may have little or no effect on preterm birth before 37 weeks of gestation (13 trials, 16 139 women; RR: 0.86, 95% CI: 0.70–1.05). However, when trials are stratified by dose (< 1000 mg vs ≥ 1000 mg), moderate-certainty evidence shows that high-dose calcium supplementation probably reduces preterm birth (12 trials, 15 479 women; RR: 0.81, 95% CI: 0.66–0.99).

Low-certainty evidence suggests that calcium supplementation may make little or no difference to perinatal mortality (8 trials, 15 785 women; RR: 0.87, 95% CI: 0.72–1.06), and moderate-certainty evidence shows that it probably has little or no effect on stillbirths or fetal deaths (6 trials, 15 269 women; RR: 0.91, 95% CI: 0.72–1.14).

Additional considerations

• In the WHO recommendations for prevention and treatment of pre-eclampsia and eclampsia (2011), the recommendation on calcium states: “In areas where dietary calcium intake is low, calcium supplementation during pregnancy (at doses of 1.5–2.0 g elemental calcium/day) is recommended for the prevention of pre-eclampsia in all women, but especially in those at high risk of developing pre-eclampsia (strong recommendation)” (57). This recommendation is based on moderate-quality evidence showing a 64% risk reduction (CI: 35–80%) in pre-eclampsia among women or populations with low baseline dietary calcium intake (57).
• In considering the evidence from the review of “non-hypertensive” effects, the GDG agreed that the effect of calcium on preterm birth is probably not distinct from the effect on preventing pre-eclampsia, as preterm birth is frequently a consequence of pre-eclampsia.

Maternal outcomes

Moderate-certainty evidence shows that vitamin A supplementation in vitamin A deficient populations during pregnancy probably reduces maternal anaemia (3 trials, 15 649 women; RR: 0.64, 95% CI: 0.43–0.94), but that it probably has little or no effect on maternal mortality (4 trials, 101 574 women; RR: 0.88, 95% CI: 0.65–1.20). Low-certainty evidence on a composite outcome for maternal infection (including fever for more than one week at one week postnatally, puerperal fever greater than 38°C, subclinical mastitis and/or bacterial vaginosis) suggests that vitamin A supplementation may reduce maternal infection (5 trials, 17 313 women; average RR: 0.45, 95% CI: 0.2–0.99). Side-effects and other maternal ANC guideline outcomes were not reported in the trials.

Fetal and neonatal outcomes

High-certainly evidence shows that vitamin A supplementation makes little or no difference to perinatal mortality (76 176 women; RR: 1.01, 95% CI: 0.95–1.07), neonatal mortality (3 trials, 89 556 neonates; RR: 0.97, 95% CI: 0.90–1.05) or stillbirths (2 trials, 122 850 neonates; RR: 1.04, 95% CI: 0.98–1.10). Moderate-certainty evidence indicates that vitamin A supplementation probably has little or no effect on low birth weight (< 2500 g) (4 trials, 14 599 neonates; RR: 0. 1.02, 95% CI: 0.89–1.16), and low-certainty evidence suggests that it may have little or no effect on preterm birth (5 trials, 40 137 women; RR: 0.98, 95% CI: 0.94–1.01). Neonatal infections and congenital anomalies were not reported in the trials.

Additional considerations

• Moderate-certainty evidence shows that vitamin A supplementation reduces night blindness in pregnant women living in areas with a high prevalence of this condition (2 trials, approximately 100 000 women; RR: 0.79, 95% CI: 0.64–0.98).
• Miscarriage and teratogenicity have been associated with high vitamin A intake within 60 days of conception; however, a WHO expert group consultation in 1998 concluded that daily doses of up to 3000 μg per day after day 60 are probably safe, especially in areas where vitamin A deficiency is common (62).

Maternal outcomes

Moderate-certainty evidence indicates that zinc supplementation probably makes little or no difference to the risk of any maternal infections (3 trials, 1185 women; RR: 1.06; 95% CI: 0.74–1.53). The evidence on caesarean section, pre-eclampsia and side-effects (maternal taste and smell dysfunction) is of very low certainty, and the review did not include anaemia, maternal mortality or maternal satisfaction as review outcomes.

Fetal and neonatal outcomes

Moderate-certainty evidence indicates that zinc supplementation probably makes little or no difference to the risk of having SGA (8 trials, 4252 newborns; RR: 1.02; 95% CI: 0.94–1.11) or low-birth-weight neonates (14 trials, 5643 neonates; RR: 0.93, 95% CI: 0.78–1.12). However, low-certainty evidence suggests that zinc supplementation may reduce preterm birth (16 trials, 7637 women; RR: 0.86, 95% CI: 0.76–0.97), particularly in women with presumed low zinc intake or poor nutrition (14 trials, 7099 women; RR: 0.87, 95% CI: 0.77–0.98).

Low-certainty evidence suggests that zinc supplementation may have little or no effect on congenital anomalies (6 trials, 1240 newborns; RR: 0.67, 95% CI: 0.33–1.34) and macrosomia (defined in the review as “high birth weight”; 5 trials, 2837 neonates; RR: 1.00, 95% CI: 0.84–1.18). Evidence on perinatal mortality and neonatal sepsis is of very low certainty.

Additional considerations

• The trials were clinically heterogeneous, therefore it is unclear what dose and timing of zinc supplementation, if any, might lead to a possible reduction in preterm birth.
• There is little or no evidence on side-effects of zinc supplementation. In addition, it is unclear to what extent zinc might compete with iron and/or calcium for absorption. Maternal anaemia was not evaluated in the review.

Maternal outcomes

High-certainty evidence shows that MMN supplementation has a similar effect to iron and folic acid supplements only (standard care) on maternal anaemia (5 trials; RR: 0.98, 95% CI: 0.85–1.13). Compared to iron and folic acid only, moderate-certainty evidence indicates that MMN supplements probably make little or no difference to caesarean section rates (4 trials; RR: 1.03, 95% CI: 0.75–1.43) and low-certainty evidence suggests that they may have little or no effect on maternal mortality (3 trials; RR: 0.97, 95% CI: 0.63–1.48). There was no evidence relating to maternal satisfaction or side-effects.

Fetal and neonatal outcomes

High-certainty evidence shows that MMN supplementation reduces the risk of having a low-birth-weight neonate compared with iron and folic acid supplements only (14 trials; RR: 0.88, 95% CI: 0.85–0.91), but moderate-certainty evidence indicates that it probably makes little or no difference to the risk of having an SGA neonate (13 trials; RR: 0.98, 95% CI: 0.96–1.00). High-certainty evidence shows that MMN supplements make little or no difference to preterm birth rates (14 trials; RR: 0.95, 95% CI: 0.88–1.03). Moderate-certainty evidence shows that MMN supplements probably make little or no difference to perinatal mortality (11 trials; RR: 1.00, 95% CI: 0.85–1.19), neonatal mortality (11 trials; RR: 0.99, 95% CI: 0.90–1.08) or stillbirths (14 trials; RR: 0.97, 95% CI: 0.86–1.09). The evidence on congenital anomalies is of low certainty and inconclusive (1 trial, 1200 women; RR: 0.99, 95% CI: 0.14–7.00).

High-certainty evidence from analyses restricted to trials of UNIMMAP only are consistent with the overall findings, with the exception that it shows that UNIMMAP reduces the risk of having an SGA neonate compared with iron and folic acid supplements only (8 trials; RR: 0.85, 95% CI: 0.77–0.94).

Subgroup analyses according to the iron dose in the control group are generally consistent with the overall findings. However, for the subgroup of studies that compared MMN supplements to 60 mg iron and 400 μg folic acid, a harmful effect of MMNs on neonatal mortality cannot be excluded (6 trials; RR: 1.22, 95% CI: 0.95–1.57).

Additional considerations

• A separate review of the effects of MMN supplementation during pregnancy on child health benefits pooled data from nine of the trials included in the Cochrane review and found no evidence of beneficial effects on child mortality, growth or cognitive function (67).

Maternal outcomes

Low-certainty evidence suggests that oral pyridoxine supplements may have little or no effect on pre-eclampsia (2 trials, 1197 women; RR: 1.71, 95% CI: 0.85–3.45). No other maternal outcomes relevant to the ANC guideline were reported in the review.

Fetal and neonatal outcomes

Trials contributed no data on low birth weight, preterm birth or other ANC guideline outcomes. Mean birth weight was evaluated in one small trial but the evidence is very uncertain. There was no evidence on congenital anomalies.

Additional considerations

• Moderate-certainty evidence shows that vitamin B6 probably provides some relief for nausea during pregnancy (see evidence summary for Recommendation D.1, in section D: Interventions for common physiological symptoms).
• Vitamin B6 deficiency alone is uncommon; it mostly occurs in combination with deficiencies of other B vitamins (70).

Maternal outcomes

Moderate-certainty evidence shows that vitamin E and C combined supplements probably have little or no effect on the risk of developing pre-eclampsia (14 studies, 20 878 women; RR: 0.91 95% CI: 0.79–1.06) and eclampsia (8 trials, 19 471 women; RR: 1.67, 95% CI: 0.82–3.41). Moderate-certainty evidence also shows that vitamin E and C supplements probably have little or no effect on maternal mortality (7 trials, 17 120 women; RR: 0.60, 95% CI: 0.14–2.51) and caesarean section (6 trials, 15 297 women; RR: 1.02, 95% CI: 0.97–1.07).

Side-effects: High-certainty evidence shows that vitamin E and C supplementation is associated with an increased risk of abdominal pain during pregnancy (1 trial, 1877 women; RR: 1.66, 95% CI: 1.16–2.37; absolute effect of 32 more per 1000 women).

Fetal and neonatal outcomes

High-certainty evidence indicates that vitamin E and C supplementation does not have an important effect on SGA (11 trials, 20 202 women; RR: 0.98, 95% CI: 0.91–1.06). Moderate-certainty evidence shows that vitamin E and C supplements probably have little or no effect on preterm birth (11 trials, 20 565 neonates; RR: 0.98, 95% CI: 0.88–1.09), neonatal infections (5 trials, 13 324 neonates; RR: 1.10, 95% CI: 0.73–1.67) and congenital anomalies (4 trials, 5511 neonates; RR: 1.16, 95% CI: 0.83–1.63).

Additional considerations

• The high-certainty evidence on abdominal pain is derived from a large, well designed trial in which abdominal pain occurred in 7.9% of women in the vitamin E and C supplement group and 4.8% of women in the placebo group.
• Despite the certainty of these effects of vitamin E and C supplementation, the biological explanations for these adverse effects are not established.
• Moderate-certainty evidence indicates that vitamin E and C supplements probably reduce the risk of placental abruption (7 trials, 14 922 women; RR: 0.64, 95% CI: 0.44–0.93; absolute effect of 3 fewer abruptions per 1000) but make little or no difference to the risk of antepartum haemorrhage from any cause (2 trials, 12 256 women; RR: 1.25, 95% CI: 0.85–1.82).
• High-certainty evidence shows vitamin E and C supplementation increases PROM at term (37 weeks of gestation or more) (2 trials, 2504 women; RR: 1.77, 95% CI: 1.37–2.28; absolute effect of 52 more cases of PROM per 1000).
• The trial contributing the most data on PROM was stopped early, based on their PROM data, when only a quarter of the planned sample (10 000 women) had been accrued.
• Low- to moderate-certainty evidence on vitamin C only suggests that vitamin C alone (in doses ranging from 100 mg to 1000 mg) may reduce preterm PROM (5 studies, 1282 women; RR: 0.66, 95% CI: 0.48–0.91) and term PROM (1 study, 170 women; RR: 0.55, 95% CI: 0.32–0.94).

Maternal outcomes

The evidence on pre-eclampsia, GDM, maternal mortality, caesarean section and side-effects is very uncertain (i.e. all findings were assessed as very low-certainty evidence).

Fetal and neonatal outcomes

Low-certainty evidence suggests that vitamin D supplementation may reduce low-birth-weight neonates (3 trials, 493 women; RR: 0.40, 95% CI: 0.24–0.67) and preterm birth (< 37 weeks of gestation) (3 trials, 477 women; RR: 0.36, 95% CI: 0.14–0.93), but may have little or no effect on neonatal deaths (2 trials, 282 women, RR: 0.27; 95% CI: 0.04–1.67) and stillbirths (3 trials, 540 women; RR: 0.35, 95% CI: 0.06–1.99).

Maternal outcomes

Moderate-certainty evidence shows that vitamin D plus calcium probably reduces pre-eclampsia (3 trials, 798 women; RR: 0.51; 95% CI: 0.32–0.80), but low-certainty evidence suggest that it may have little or no effect on GDM (1 trial, 54 women, 1 event; RR: 0.43, 95% CI: 0.05–3.45).

Fetal and neonatal outcomes

Moderate-certainty evidence indicates that vitamin D plus calcium probably increases preterm birth (< 37 weeks of gestation) (3 trials, 798 women; RR: 1.57, 95% CI: 1.02–2.43). Low-certainty evidence suggests that vitamin D plus calcium has little or no effect on neonatal mortality (1 trial, 660 women; RR: 0.20, 95% CI: 0.01–4.14).

Additional considerations

• Due to the limited evidence currently available to directly assess the benefits and harms of the use of vitamin D supplementation alone in pregnancy for improving maternal and infant health outcomes, the use of this intervention during pregnancy as part of routine ANC is not recommended (75).
• The moderate-certainty evidence showing that adding vitamin D to calcium supplementation probably increases preterm birth is of concern and this potential harm needs further investigation.

Maternal outcomes

None of the maternal outcomes addressed in the ANC guideline were reported in the review.

Fetal and neonatal outcomes

Low-certainty evidence from one trial shows that restricting caffeine intake (replacing caffeinated coffee with decaffeinated coffee) may have little or no effect on SGA (1150 neonates; RR: 0.97, 95% CI: 0.57–1.64), mean birth weight (1197 neonates; MD: 20.00, 95% CI: −48.68 to 88.68) and preterm birth (1153 neonates; RR: 0.81, 95% CI: 0.48–1.37).

No data were available on congenital anomalies or perinatal mortality.

Fetal and neonatal outcomes: low birth weight

Moderate-certainty evidence from one review shows that high caffeine intake (more than 300 mg) is probably associated with a greater risk of low birth weight than low or no caffeine intake (12 studies; odds ratio [OR]: 1.38, 95% CI: 1.10–1.73) (78). Very low- to moderate-certainty evidence from the other review was stratified according to dose and shows that very low caffeine intake may be associated with fewer low-birth-weight neonates than low (5 studies; RR: 1.13, 95% CI: 1.06–1.21), moderate (7 studies; RR: 1.38, 95% CI: 1.18–1.62) or high caffeine intake (8 studies; RR: 1.60, 95% CI: 1.24–2.08) (77).

Fetal and neonatal outcomes: stillbirths

The reviews reported “pregnancy loss”, a composite outcome comprising stillbirths and miscarriages. Moderate-certainty evidence from one review (80) shows that any caffeine intake probably increases pregnancy loss compared with controls (no exposure) (18 studies; OR: 1.32, 95% CI: 1.24–1.40). However, pregnancy loss is probably more common among pregnant women with moderate caffeine intake (18 studies; OR: 1.28, 95% CI: 1.16–1.42) and high caffeine intake (17 studies, OR: 1.60, 1.46–1.76), but not more common with low caffeine intake (13 studies; OR: 1.04, 95% CI: 0.94–1.15) compared with controls. This NRS evidence was upgraded to “moderate-certainty” due to the presence of a dose–response relationship. A dose–response relationship was also observed in the other review but the evidence was less certain (79).

Test accuracy of on-site Hb testing with haemoglobinometer and haemoglobin colour scale (HCS) methods to detect anaemia (EB Table B.1.1)

The evidence was derived from a test accuracy review conducted to support the ANC guideline (81). Only one study (671 women) contributed data (87). The study, conducted in Malawi, assessed the test accuracy of on-site Hb testing with a haemoglobinometer (HemoCue^®) and the HCS method in comparison to a full blood count test performed by an electronic counter (Coulter counter), the reference standard.

Moderate-certainty evidence shows that the sensitivity and specificity of the haemoglobinometer test in detecting anaemia (Hb < 110 g/L) are approximately 0.85 (95% CI: 0.79–0.90) and 0.80 (95% CI: 0.76–0.83), respectively, while the sensitivity and specificity of the HCS method are lower at approximately 0.75 (95% CI: 0.71–0.80) and 0.47 (95% CI: 0.41–0.53), respectively.

For severe anaemia (defined in the study as Hb < 60 g/L), moderate-certainty evidence shows that the sensitivity and specificity of the haemoglobinometer test are approximately 0.83 (95% CI: 0.44–0.97) and 0.99 (95% CI: 0.98–1.00), respectively, while for the HCS method they are approximately 0.50 (95% CI: 0.15–0.85) and 0.98 (95% CI: 0.97–0.99), respectively.

Values

Please see “Women’s values” in section 3.B.1: Maternal assessment: Background (p. 41).

Resources

Any health-care provider can perform both the haemoglobinometer and HCS methods after minimal training. The haemoglobinometer and HCS methods have been estimated to cost approximately US$ 0.75 and US$ 0.12 per test, respectively (82). Both methods require needles for finger pricks, cotton balls, gloves and Sterets^® skin cleansing swabs; however, the higher costs associated with haemoglobinometer tests are mainly due to supplies (cuvettes and controls), equipment costs and maintenance.

Equity

The highest prevalence of maternal anaemia occurs in Africa and South-East Asia, where parasitic infections are major contributory factors (33). Anaemia increases perinatal risks for mothers and newborns and contributes to preventable mortality. Accurate, low-cost, simple-to-use tests to detect anaemia might help to address health inequalities by improving the detection and subsequent management of women with anaemia, particularly severe anaemia, in low-resource settings.

Acceptability

Qualitative evidence from a variety of settings indicates that women generally appreciate clinical tests that support their well-being during pregnancy (moderate confidence in the evidence) (22). However, evidence from LMICs indicates that where there are likely to be additional costs associated with tests, or where the recommended interventions are unavailable because of resource constraints, women may be less likely to engage with ANC services (high confidence in the evidence).

Feasibility

Qualitative evidence from providers in various LMICs indicates that a lack of resources, both in terms of the availability of the diagnostic equipment and potential treatments, as well as the lack of suitably trained staff to deliver the service, may limit implementation of recommended interventions (high confidence in the evidence) (45).

Test accuracy of on-site urine Gram staining and dipsticks to detect ASB (EB Table B.1.2)

The evidence was derived from a test accuracy review of on-site urine tests conducted to support the ANC guideline (88). Four studies (1904 pregnant women) contributed data on urine Gram staining and eight studies (5690 pregnant women) contributed data on urine dipsticks. Most of the studies were conducted in LMICs. The average prevalence of ASB in the studies was 8%. A Gram stain was positive if one or more bacteria were detected per oil-immersed field, and a dipstick test was positive if it detected either nitrites or leucocytes. The reference standard used was urine culture with a threshold of 10⁵ cfu/mL.

However, the certainty of the evidence on the accuracy of both Gram stain tests and dipstick tests is very low, with pooled sensitivity and specificity of the Gram stain test estimated at 0.86 (95% CI: 0.80–0.91) and 0.97 (95% CI: 0.93–0.99), respectively, and pooled sensitivity and specificity for urine dipsticks estimated at 0.73 (95% CI: 0.59–0.83) and 0.89 (95% CI: 0.79–0.94), respectively. A positive nitrite test alone on dipsticks was found to be less sensitive but more specific than when urine leucocytes were also considered.

Values

Please see “Women’s values” in section 3.B.1: Maternal assessment: Background (p. 41).

Resources

Dipsticks are relatively low cost compared with the Gram stain test, as the latter requires trained staff and laboratory equipment and supplies (microscope, glass slides, reagents, Bunsen burner or slide warmer). Gram stain tests take longer to perform and to produce results than urine dipstick tests (10–30 minutes vs 60 seconds).

Equity

Preterm birth is the leading cause of neonatal death worldwide, with most deaths occurring in LMICs. Timely diagnosis and treatment of risk factors associated with preterm birth might therefore help to address health inequalities.

Acceptability

Qualitative evidence from a range of settings suggests that women view ANC as a source of knowledge, information and clinical expertise and that they generally appreciate the tests and advice they are offered (high confidence in the evidence) (22). However, engagement with ANC services may be limited if tests and procedures are not explained properly or when women feel their beliefs and traditions are being overlooked or ignored by health-care professionals. In addition, if the Gram stain test is associated with long waiting times at ANC or having to return for test results, this may be less acceptable to women, as it might have additional cost and convenience implications for them (high confidence in the evidence). Health professionals are likely to prefer the dipstick test as it is associated with less effort (no need to label samples for laboratory assessment, perform tests or schedule follow-up visits to provide the results) and might provide additional information pertaining to other conditions (pre-eclampsia and diabetes mellitus) (high confidence in the evidence).

Feasibility

Qualitative evidence indicates that, in some LMIC settings, the lack of diagnostic equipment at ANC facilities discourages women from attending, and that providers often do not have the diagnostic equipment, supplies or skills to perform tests (high confidence in the evidence) (45). Therefore, urine dipstick tests, which are cheaper and easy to perform, might be more feasible in low-resource settings.

Effects of universal screening to detect IPV compared with no screening (usual care) (EB Table B.1.3)

The evidence on screening for IPV was derived from a Cochrane review that included two trials conducted in urban ANC settings in HICs (Canada and the USA), involving 663 pregnant women (89). In one trial, 410 women were randomized before 26 weeks of gestation to a computer-based abuse assessment screening tool, with and without a provider cue sheet (giving the results of the assessment to the provider), prior to ANC consultation with a health-care provider. In the other trial (a cluster-RCT), providers administered a face-to-face screening tool that screened for 15 risk factors, including IPV, to women between 12 and 30 weeks of gestation in the intervention clusters, while women in the control clusters received usual ANC.

Low-certainty evidence from the review suggests that abuse assessment screening may identify more pregnant women with IPV than those identified through usual ANC (2 trials, 663 women; OR: 4.28, 95% CI: 1.77–10.36).

Values

Please see “Women’s values” in section 3.B.1: Maternal assessment: Background (p. 41).

Resources

Clinical enquiry about IPV can be conducted face-to-face or by providing women with a written or computer-based questionnaire. Although the costs of implementing these methods can vary, they might be relatively low. Subsequent management and IPV support linked to the screening intervention, however, requires sophisticated training and can therefore have significant cost implications. The GDG considered that training and resources in low-resource settings might be best targeted towards first response to IPV rather than IPV screening.

Equity

IPV is highly prevalent in many LMICs and among disadvantaged populations (92, 93). Effective interventions to enquire about IPV in disadvantaged populations might help to identify those at risk of IPV-related adverse outcomes, and facilitate the provision of appropriate supportive interventions leading to improved equity. However, more evidence is needed.

Acceptability

Qualitative evidence from a range of settings on women’s views of ANC suggests that pregnant women would like to be seen by a kind and supportive health-care provider who has the time to discuss issues of this nature in a private setting (high confidence in the evidence) (22). However, evidence from LMICs suggests that women may be unlikely to respond favourably to cursory exchanges of information with providers who they sometimes perceive to be hurried, uncaring and occasionally abusive (high confidence in the evidence). In addition, some women may not appreciate enquiries of this nature, particularly those living in male-dominated, patriarchal societies, where women’s financial dependence on their husbands may influence their willingness to discuss IPV, especially if the health professional is male (22).

From the providers’ perspective, qualitative evidence mainly from HICs suggests that providers often find it difficult to enquire about for IPV for the following reasons: they do not feel they have enough knowledge, training or time to discuss IPV in a sensitive manner; the presence of the partner acts as a barrier; they may have experienced IPV themselves; and they lack knowledge and guidance about the availability of additional support services (counselling, social work, etc.) (high confidence in the evidence). Providers highlight the midwife-led continuity of care (MLCC) model as a way of achieving a positive, trusting and empathetic relationship with pregnant women (moderate confidence in the evidence) (see Recommendation E.2, in section E: Health systems interventions to improve the utilization and quality of ANC).

Feasibility

Following IPV clinical enquiry, complex, multifaceted, culturally specific interventions are required to manage IPV, which could be challenging in many low-resource settings. However, emerging evidence from HICs shows that medium-duration empowerment counselling and advocacy/support, including a safety component, offered by trained health-care providers could be beneficial, and the feasibility of such interventions in LMIC settings needs investigation (86).

Effects of daily maternal fetal movement counting compared with standard ANC (EB Table B.2.1)

The evidence on the effects of daily fetal movement counting was derived from a Cochrane review (107). Two RCTs from HICs contributed data for this comparison. One was a large, multicentre, cluster RCT (68 654 women) conducted in Belgium, Ireland, Sweden, the United Kingdom and the USA, which compared a “count-to-ten” fetal movement counting kick chart with standard ANC in women with uncomplicated pregnancies recruited between 28 and 32 weeks of gestation. Women in the standard ANC group were asked about fetal movements at each ANC visit. The other trial was a multicentre RCT conducted in Norway involving 1123 women that compared a modified “count-to-ten” fetal movement counting protocol with standard care.

Values

Please see “Women’s values” in section 3.B.2: Fetal assessment: Background (p. 54).

Resources

Fetal movement counting is a low-cost intervention on its own, but it could be resource-intensive if it leads to unnecessary additional interventions or hospital admissions.

Equity

LMICs bear the global burden of perinatal morbidity and mortality, and women who are poor, least educated and residing in rural areas of LMICs have lower ANC coverage and worse pregnancy outcomes than more advantaged women (29). Therefore, simple, effective, low-cost antenatal interventions to assess fetal well-being could help to address health inequalities by improving detection of complications in low-resource settings.

Acceptability

Qualitative evidence shows that women generally appreciate the knowledge and information they can acquire from health-care providers during ANC visits, provided this is explained properly and delivered in a consistent, caring and culturally sensitive manner (high confidence in the evidence) (22). It also shows that health professionals want to give appropriate information and advice to women but sometimes they don’t feel suitably trained to do so (high confidence in the evidence) (45).

Feasibility

From the perspective of women who live far from ANC clinics and who may not have the resources or time to attend ANC regularly, and the perspective of ANC providers with limited resources, this intervention may offer a practical and cost–effective approach to monitoring fetal well-being if it’s shown to be effective (high confidence in the evidence) (22, 45).

Effects of SFH measurement versus abdominal palpation (EB Table B.2.2)

The evidence on the effects of SFH measurement was derived from a Cochrane review that included only one trial conducted in Denmark involving 1639 pregnant women enrolled at about 14 weeks of gestation (109). SFH measurement or abdominal palpation were performed from 28 weeks of gestation. Most women had at least three assessments, with measurements plotted on a chart.

Values

Please see “Women’s values” in section 3.B.2: Fetal assessment: Background (p. 54).

Resources

Both abdominal palpation and SFH measurement are low-cost interventions with the main cost being staff training. SFH requires tape measures to be available.

Equity

LMICs bear the global burden of perinatal morbidity and mortality, and women who are poor, least educated and residing in rural areas of LMICs have lower ANC coverage and worse pregnancy outcomes than more advantaged women (29). Therefore, simple, effective, low-cost, routine antenatal interventions to assess fetal well-being could help to address health inequalities by improving detection of complications in low-resource settings.

Acceptability

SFH and clinical palpation are non-invasive approaches for fetal assessment, which are widely used and not known to be associated with acceptability issues. However, in some settings women experience a sense of shame during physical examinations, and this needs to be addressed with sensitivity by health-care providers (low confidence in the evidence) (22).

Feasibility

Both methods are considered equally feasible, provided tape measures are available.

Effects of routine antenatal CTG versus no routine antenatal CTG (EB Table B.2.3)

A Cochrane review of routine antenatal CTG for fetal assessment identified no eligible studies of routine CTG and all six included studies involved women with high-risk pregnancies (117).

Values

Please see “Women’s values” in section 3.B.2: Fetal assessment: Background (p. 54).

Resources

CTG machines are costly (starting from about US$ 450)⁴, require maintenance and supplies of ultrasound gel, and require staff training in their use and interpretation.

Equity

Simple, effective, low-cost, antenatal interventions to assess fetal well-being could help to address health inequalities by improving detection of complications in low-resource settings, which bear the burden of perinatal mortality.

Acceptability

Qualitative evidence from a variety of settings indicates that women generally appreciate the use of technology to monitor pregnancy (high confidence in the evidence), and a lack of modern equipment at ANC facilities in LMICs may discourage women from attending (moderate confidence in the evidence) (22). However, in some LMICs, women hold the belief that pregnancy is a healthy condition and may be resistant to CTG use unless they have experienced a previous pregnancy complication (high confidence in the evidence). Acceptability may be further compromised if the reasons for using CTG are not properly explained (high confidence in the evidence).

Feasibility

Health-care providers in LMIC settings feel that a lack of modern equipment and training limits the implementation of this type of intervention (high confidence in the evidence) (45).

a. Effects of an ultrasound scan before 24 weeks of gestation (early ultrasound scan) versus selective ultrasound scan (EB Table B.2.4a)

The evidence on early ultrasound was derived from a Cochrane review that included 11 RCTs conducted in Australia, Norway, South Africa, Sweden, the United Kingdom and the USA, involving 37 505 women (120). The intervention in all trials involved an ultrasound scan before 24 weeks of gestation, with women in the control arm undergoing selective scans if indicated (or, in one study, concealed scans, the results of which were not shared with clinicians unless requested). The scans usually included assessment of gestational age (biparietal diameter with or without head circumference and femur length), fetal anatomy, number of fetuses and location of the placenta. Scans were performed in most trials between 10 and 20 weeks of gestation, with three trials evaluating scans before 14 weeks, and three trials evaluating an intervention comprising both early (at 18–20 weeks) and late scans (at 31–33 weeks).

b. Effects of an ultrasound scan after 24 weeks of gestation (late ultrasound scan) versus no late ultrasound scan (EB Table B.2.4b)

This evidence on late ultrasound was derived from a Cochrane review that included 13 RCTs conducted in HICs (121). Most women in these trials underwent early ultrasound scan and were randomized to receive an additional third trimester scan or to selective or concealed ultrasound scan. The purpose of the late scan in these trials, which was usually performed between 30 and 36 weeks of gestation, variably included assessment of fetal anatomy, estimated weight, amniotic fluid volume and/or placental maturity.

Values

Please see “Women’s values” in section 3.B.2: Fetal assessment: Background (p. 54).

Resources

The cost of ultrasound equipment, especially portable compact units, has decreased (122), and they are currently available at less than US$ 10 000 (28). Thus, given the cost of equipment, maintenance, supplies (ultrasound gel), replacement batteries, initial and ongoing staff training and supervision, and staffing costs (allowing 15–45 minutes per scan), routine ultrasound scans may have considerable resource implications for LMIC settings.

Equity

Effective interventions to increase uptake and quality of ANC services, and improve the experience of care, are needed in LMICs to prevent maternal and perinatal mortality and improve equity. However, if women are expected to pay for ultrasound scans, or if scans are not available to women living in rural areas due to feasibility issues, this intervention could perpetuate inequalities. In addition, ultrasound sexing of the fetus in some low-income countries has a negative impact on gender equity and needs to be monitored.

Acceptability

Qualitative evidence shows that women generally appreciate the knowledge and information they can acquire from health-care providers and that they are willing to be screened and tested for a variety of conditions, provided the information and procedures are explained properly and delivered in a caring and culturally sensitive manner (high confidence in the evidence) (22). Evidence also shows that, in some LMICs, the lack of modern technology (like ultrasound equipment) at ANC facilities discourages some women from attending (high confidence in the evidence) (22). This suggests that the offer of ultrasound might attract women to use ANC facilities, which may also lead to earlier ANC attendance. Specific studies not included in the main qualitative review indicate that women value the opportunity to see their baby via ultrasound and find the test reassuring (123). However, there is some evidence that women do not understand that ultrasound is a diagnostic tool, and that adverse findings during scans might increase anxiety and distress (124).

Qualitative evidence from health-care providers shows that they generally want to provide screening and testing procedures, but sometimes don’t feel suitably trained to do so (high confidence in the evidence) (45). This suggests that they might welcome ultrasound scans to assist with accurate gestational age estimation and to identify potential risk factors, such as multiple pregnancies, if appropriately trained and supported.

Feasibility

Feasibility challenges of antenatal ultrasound scans in LMICs includes equipment procurement and staff training, ensuring a power supply (via a power point or rechargeable batteries) and secure storage, regular equipment maintenance, maintaining adequate and continual supplies of ultrasound gel, and ongoing technical support and supervision.

Effects of Doppler ultrasound examination of fetal blood vessels compared with no Doppler ultrasound examination (EB Table B.2.5)

The evidence on Doppler ultrasound examination was derived from a Cochrane review that included five trials involving 14 624 women in HICs (Australia, France and the United Kingdom) (114). One study evaluated a single Doppler examination at 28–34 weeks of gestation, three studies evaluated multiple Doppler examinations from as early as 18 weeks, and one study evaluated women undergoing single or multiple examinations from 26 to 36 weeks of gestation. Data were evaluated together and separately for single and multiple examinations. Women in the control arms received standard ANC with no (or concealed) Doppler examination.

Values

Please see “Women’s values” in section 3.B.2: Fetal assessment: Background (p. 54).

Resources

The cost of ultrasound equipment, especially portable compact units, has decreased (122), and they are currently available at less than US$ 10 000 (28). Thus, given the cost of equipment, maintenance, supplies (ultrasound gel), replacement batteries, initial and ongoing staff training and supervision, and staffing costs, routine Doppler ultrasound scans may have considerable resource implications for LMIC settings.

Equity

RCT evidence on maternal and perinatal effects of Doppler ultrasound examination is currently derived from HICs and high-quality research is needed on this intervention in LMICs to determine whether, by improving detection of pregnancy complications, it can reduce perinatal mortality and improve health equity.

Acceptability

Qualitative evidence shows that women generally appreciate the knowledge and information they can acquire from health-care providers and that they are willing to be screened and tested for a variety of conditions, provided the information and procedures are explained properly and delivered in a caring and culturally sensitive manner (high confidence in the evidence) (22). Evidence also shows that, in some LMICs, the lack of modern technology (like ultrasound equipment) at ANC facilities discourages some women from attending (high confidence in the evidence) (22).

Qualitative evidence from health-care providers shows that they generally want to provide screening and testing procedures, but sometimes don’t feel suitably trained to do so (high confidence in the evidence) (45). This suggests that they might welcome Doppler ultrasound scans to identify potential risk factors, if appropriately trained and supported.

Feasibility

Feasibility challenges of Doppler ultrasound scans in LMICs include equipment procurement and staff training, ensuring a power supply (via a power point or rechargeable batteries) and secure storage, regular equipment maintenance, maintaining adequate and continual supplies of ultrasound gel, and ongoing technical support and supervision.

Maternal outcomes

The only maternal ANC guideline outcomes reported were infection outcomes. Low-certainty evidence suggests that antibiotics may reduce persistent bacteriuria (4 trials, 596 women; RR: 0.30, 95% CI: 0.18–0.53); however, the evidence on the effect on pyelonephritis is very uncertain.

Fetal and neonatal outcomes

Low-certainty evidence suggests that antibiotics for ASB may reduce low-birth-weight neonates (8 trials, 1437 neonates; RR: 0.64, 95% CI: 0.45–0.93) and preterm birth (2 trials, 142 women; RR: 0.27, 95% CI: 0.11–0.62). No other ANC guideline outcomes were reported.

Additional considerations

• The GDG also evaluated evidence on treatment duration (single dose versus short-course [4–7 days]) from a related Cochrane review that included 13 trials involving 1622 women (136). Ten trials compared different durations of treatment with the same antibiotic, and the remaining three compared different durations of treatment with different drugs. A wide variety of antibiotics was used. The resulting pooled evidence on bacterial persistence (7 trials), recurrent ASB (8 trials) and pyelonephritis (2 trials) was judged as very uncertain. However, on sensitivity analysis including high-quality trials of amoxicillin and nitrofurantoin only, the high-certainty evidence indicates that bacterial persistence is reduced with a short course rather than a single dose (2 trials, 803 women; RR: 1.72, 95% CI: 1.27–2.33). High-certainty evidence from one large trial shows that a seven-day course of nitrofurantoin is more effective than a one-day treatment to reduce low birth weight (714 neonates; RR: 1.65, 95% CI: 1.06–2.57). Low-certainty evidence suggests that single-dose treatments may be associated with fewer side-effects (7 trials, 1460 women; RR: 0.70, 95% CI: 0.56–0.88). See Web supplement (EB Table C.1).
• The GDG also evaluated evidence on the test accuracy of urine Gram staining and dipstick testing (see Recommendation B.1.2 in section 3.B).

Maternal outcomes

Evidence from this single study on the risk of recurrent pyelonephritis and RUTI with prophylactic antibiotics is very uncertain. No other maternal ANC guideline outcomes were reported in the study.

Fetal and neonatal outcomes

Evidence on the risk of low birth weight and preterm birth with prophylactic antibiotics is very uncertain. No other fetal and neonatal ANC guideline outcomes were reported in the study.

Additional considerations

• Antibiotic prophylaxis to prevent RUTI may lead to increased antimicrobial resistance and there is a lack of evidence on this potential consequence.

Fetal and neonatal outcomes

Evidence on the effect of antenatal anti-D on RhD alloimmunization during pregnancy, suggesting little or no difference in effect, is very uncertain. In addition, the evidence on the effect on postpartum RhD alloimmunization and alloimmunization up to 12 months postpartum among women giving birth to Rh-positive newborns (n = 2297 and 2048, respectively) is very uncertain, partly because events were rare. Evidence on the effect of antenatal anti-D on neonatal morbidity (jaundice) from one trial (1882 neonates) is also very uncertain, partly because events were rare. No other ANC guideline outcomes were reported in the review.

Additional considerations

• Low-certainty evidence from the Cochrane review suggests that Rh-negative women who receive antenatal anti-D are less likely to register a positive Kleihauer test (which detects fetal cells in maternal blood) during pregnancy (1 trial, 1884 women; RR: 0.60, 95% CI: 0.41–0.88) and at the birth of a Rh-positive neonate (1 trial, 1189 women; RR: 0.60, 95% CI: 0.46–0.79).
• In the Cochrane review, the rate of RhD alloimmunization during pregnancy, the postpartum period and up to 12 months later among women in the control group was 0.6%, 1.1% and 1.5%, respectively.
• Rates of RhD alloimmunization in subsequent pregnancies were not reported in the trials.
• There is no evidence on optimal dose of antenatal anti-D prophylaxis and various regimens are used. There are two ongoing studies listed in the Cochrane review, which may help to clarify issues around effects and dosage once completed.
• Only 60% of Rh-negative primigravidas will have an Rh-positive newborn, therefore 40% of Rh-negative women will receive anti-D unnecessarily with antenatal anti-D prophylaxis (138).

Maternal outcomes

Low-certainty evidence suggests that a single dose of albendazole or mebendazole in the second trimester of pregnancy may have little or no effect on maternal anaemia (defined as Hb < 11 g/dL) (4 trials, 3266 women; RR: 0.94; 95% CI: 0.81–1.10).

Fetal and neonatal outcomes

Moderate-certainty evidence indicates that a single dose of albendazole or mebendazole in the second trimester of pregnancy probably has little or no effect on preterm birth (2 trials, 1318 women; RR: 0.88, 95% CI: 0.43–1.78) or perinatal mortality (2 trials, 3385 women; RR: 1.09, 95% CI: 0.71–1.67). No other ANC guideline outcomes were reported in the review.

Additional considerations

• None of the trials in the Cochrane review evaluated effects of more than one dose of anthelminthics. Findings from large non-randomized studies (NRSs) suggest that prophylactic anthelminthic treatment may have beneficial effects for mothers and newborns living in endemic areas (143–145):

  –

  One NRS, including approximately 5000 pregnant women in Nepal with a 74% prevalence of hookworm infection, reported a 41% reduction in six-month infant mortality among women receiving two doses of albendazole (one each in the second and third trimesters) compared with no treatment (95% CI: 18–57%) (143). This study also showed reductions in severe maternal anaemia with albendazole.

  –

  A study from Sri Lanka involving approximately 7000 women compared mebendazole with no treatment and found fewer stillbirths and perinatal deaths among women receiving mebendazole (1.9% vs 3.3%; OR: 0.55, 95% CI: 0.40–0.77), and little difference in the occurrence of congenital anomalies (1.8% vs 1.5%, for intervention and controls, respectively; OR: 1.24, 95% CI: 0.80–1.91), even among the 407 women who had taken mebendazole in the first trimester against medical advice (145).

• The WHO manual on Preventive chemotherapy in human helminthiasis stresses that every opportunity should be taken to reach at-risk populations through existing channels (141).
• Cross-referencing other WHO guidelines, the upcoming 2016 WHO Guideline: preventive chemotherapy to control soil-transmitted helminth infections in high-risk groups recommends that a single dose of albendazole or mebendazole should be offered to pregnant women in the second and third trimesters of pregnancy where the prevalence of any soil-transmitted helminth infection (roundworm, hookworm and whipworm) exceeds 20% (140).
• Preventive helminthic treatment helps to lessen the burden of other infections, e.g. HIV, malaria and TB, and contributes to a sustained reduction of transmission (142).

Maternal outcomes

Low-certainty evidence suggests that local side-effects, such as pain, were more common with the Tdap vaccination than placebo (48 women; RR: 3.94, 95% CI: 1.41–11.01). There is no evidence on other maternal outcomes.

Fetal and neonatal outcomes

Low-certainty evidence from the Colombian trial suggests that there may be fewer neonatal tetanus cases among neonates whose mothers receive TT vaccination than among those who do not (1182 neonates; RR: 0.20, 95% CI: 0.10–0.40). Moderate-certainty evidence suggests that two or more doses of TT probably reduce neonatal mortality from any cause (1 trial, 688 neonates; RR: 0.31, 95% CI: 0.17–0.55). Further low-certainty evidence suggests that neonatal mortality from tetanus may be reduced among neonates whose mothers receive at least two TT doses (1 trial, 688 neonates; RR: 0.02, 95% CI: 0.00–0.30), but not among neonates whose mothers receive only one dose (1 trial, 494 neonates; RR: 0.57, 95% CI: 0.26–1.24). Congenital anomalies and other ANC guideline outcomes were not reported in the trials.

Additional considerations

• A systematic review that pooled data from the Colombian trial with that of a large cohort study of antenatal TT vaccination from India found moderate-certainty evidence to support a large effect (94% reduction) on neonatal tetanus deaths in favour of TT vaccination with at least two doses in pregnant women and women of childbearing age (2 trials, 2146 neonates; RR: 0.06, 95% CI: 0.02–0.20) (149).
• TT vaccination has been widely used over 40 years, leading to a substantial decrease in neonatal tetanus and an increase in neonatal survival, with no sign of possible harm to pregnant women or their fetuses (150). The WHO strategy for eliminating maternal and neonatal tetanus includes immunization of pregnant women, supplementary immunization activities in selected high-risk areas, promotion of clean deliveries and clean cord practices, and reliable neonatal tetanus surveillance (134).

Non-pharmacological agents versus placebo or no treatment

Ten trials evaluated non-pharmacological interventions including ginger (prepared as syrup, capsules or powder within biscuits) (7 trials from the Islamic Republic of Iran, Pakistan, Thailand and the USA involving 578 participants), lemon oil (one Iranian study, 100 participants), mint oil (one Iranian study, 60 participants), chamomile (one Iranian study, 105 participants), and vitamin B6 interventions (two studies in Thailand and the USA; 416 participants) compared with no treatment or placebo.

Ginger: Low-certainty evidence from several small individual studies suggests that ginger may relieve symptoms of nausea and vomiting. A study from Pakistan found that ginger reduced nausea symptom scores (68 women; MD: 1.38 lower on day 3, 95% CI: 0.03–2.73 lower), and vomiting symptom scores (64 women; MD: 1.14 lower, 95% CI: 0.37–1.91 lower), and an Iranian study showed improvements in nausea and vomiting symptom scores on day 7 in women taking ginger supplements compared with placebo (95 women; MD: 4.19 lower, 95% CI: 1.73–6.65 lower). Data from the studies in Thailand and the USA showed a similar direction of effect on nausea symptoms in favour of ginger.

Lemon oil: Low-certainty evidence from one small Iranian study suggests that lemon oil may make little or no difference to nausea and vomiting symptom scores (100 women; MD: 0.46 lower on day 3, 95% CI: 1.27 lower to 0.35 higher), or to maternal satisfaction (the number of women satisfied with treatment) (1 trial, 100 women; RR: 1.47, 95% CI: 0.91–2.37).

Mint oil: The evidence on mint oil’s ability to relieve symptoms of nausea and vomiting is of very low certainty.

Chamomile: Low-certainty evidence from one small study suggests that chamomile may reduce nausea and vomiting symptoms scores (70 women; MD: 5.74 lower, 95% CI: 3.17–8.31 lower).

Vitamin B6 (pyridoxine): Moderate-certainty evidence from two trials (one used 25 mg oral vitamin B6 8-hourly for 3 days, the other used 10 mg oral vitamin B6 8-hourly for 5 days) shows that vitamin B6 probably reduces nausea symptoms scores (388 women, trials measured the change in nausea scores from baseline to day 3; MD: 0.92 higher score change, 95% 0.4–1.44 higher), but low-certainty evidence suggests that it may have little or no effect on vomiting (2 trials, 392 women; RR: 0.76, 95% CI: 0.35–1.66).

Acupuncture and acupressure versus placebo or no treatment

Five studies (601 participants) evaluated P6 (inner forearm) acupressure versus placebo, one Thai study (91 participants) evaluated auricular acupressure (round magnetic balls used as ear pellets) versus no treatment, one study in the USA (230 participants) evaluated P6 acustimulation therapy (nerve stimulation at the P6 acupuncture point) versus placebo, and a four-arm Australian study (593 women) evaluated traditional Chinese acupuncture or P6 acupuncture versus P6 placebo acupuncture or no intervention.

Low-certainty evidence suggests that P6 acupressure may reduce nausea symptom scores (100 women; MD: 1.7 lower, 95% CI: 0.99–2.41 lower) and reduce the number of vomiting episodes (MD: 0.9 lower, 95% CI: 0.74–1.06 lower). Low-certainty evidence suggests that auricular acupressure may also reduce nausea symptom scores (91 women; MD: 3.6 lower, 95% CI: 0.58–6.62 lower), as may traditional Chinese acupuncture (296 women; MD: 0.7 lower, 95% CI: 0.04–1.36 lower). Low-certainty evidence suggests that P6 acupuncture may make little or no difference to mean nausea scores compared with P6 placebo acupuncture (296 women; MD: 0.3 lower, 95% CI: 1.0 lower to 0.4 higher).

Pharmacological agents versus placebo

One study evaluated an antihistamine (doxylamine) and another evaluated a dopamine-receptor antagonist (metoclopramide). Certain other drugs evaluated in the review (hydroxyzine, thiethylperazine and fluphenazine) are from old studies and these drugs are no longer used in pregnant women due to safety concerns.

Moderate-certainty evidence suggests that doxylamine plus vitamin B6 probably reduces nausea and vomiting symptom scores compared with placebo (1 study, 256 women; MD: 0.9 lower on day 15, 95% CI: 0.25–1.55 lower). Low-certainty evidence from this study suggests that there may be little or no difference in headache (256 women; RR: 0.81, 95% CI: 0.45–1.48) or drowsiness (256 women; RR: 1.21, 95% CI: 0.64–2.27) between doxylamine plus vitamin B6 and placebo.

Low-certainty evidence on metoclopramide (10 mg) suggests that this agent may reduce nausea symptom scores (1 trial, 68 women; MD: 2.94 lower on day 3, 95% CI: 1.33–4.55 lower). There was no side-effect data on metoclopramide in the review.

No studies compared ondansetron (a 5HT3 receptor antagonist) with placebo. Two small studies compared ondansetron with metoclopramide and doxylamine, respectively, but evidence on relative effects was uncertain.

Additional considerations

• Low-certainty evidence from single studies comparing different non-pharmacological interventions with each other – namely acupuncture plus vitamin B6 versus P6 acupuncture plus placebo (66 participants), traditional acupuncture and P6 acupuncture (296 participants), ginger versus chamomile (70 participants), P6 acupuncture versus ginger (98 participants), and ginger versus vitamin B6 (123 participants) – suggests there may be little or no difference in effects on relief of nausea symptoms.
• Low-certainty evidence suggests that there may be little or no difference between ginger and metoclopramide on nausea symptom scores (1 trial, 68 women; MD: 1.56 higher, 95% 0.22 lower to 3.34 higher) or vomiting symptom scores (68 women; MD: 0.33 higher, 95% CI: 0.69 lower to 1.35 higher) on day 3 after the intervention.
• Side-effects and safety of pharmacological agents were poorly reported in the included studies. However, drowsiness is a common side-effect of various antihistamines used to treat nausea and vomiting.
• Metoclopramide is generally not recommended in the first trimester of pregnancy, but is widely used (163). A study of over 81 700 singleton births in Israel reported that they found no statistically significant differences in the risk of major congenital malformations, low birth weight, preterm birth or perinatal death between neonates exposed (3458 neonates) and not exposed to metoclopramide in the first trimester of gestation.

Pharmacological interventions versus placebo

Low-certainty evidence suggests that complete relief from heartburn may occur more frequently with magnesium hydroxide–aluminium hydroxide–simethicone liquid and tablets than placebo (156 women; RR: 2.04, 95% CI: 1.44–2.89).

Pharmacological interventions versus advice on diet and lifestyle changes

Low-certainty evidence suggests that complete relief from heartburn may occur more frequently with sucralfate than with advice on diet and lifestyle changes (65 women; RR: 2.41, 95% CI: 1.42–4.07).

Acupuncture versus no treatment

Data on relief of heartburn was not available in the review for this comparison. Low-certainty evidence suggests that weekly acupuncture in pregnant women with heartburn may improve the ability to sleep (36 women; RR: 2.80, 95% CI: 1.14–6.86) and eat (36 women; RR: 2.40, 95% CI: 1.11–5.18), a proxy outcome for maternal satisfaction.

Additional considerations

• Heartburn during pregnancy is a common problem that can be self-treated with over-the-counter products containing antacids such as magnesium carbonate, aluminium hydroxide or calcium carbonate.
• The Cochrane review found no evidence on prescription drugs for heartburn, such as omeprazole and ranitidine, which are not known to be harmful in pregnancy (159).

Oral magnesium versus placebo

In three small studies, women in the intervention group were given 300–360 mg magnesium per day in two or three divided doses. Studies measured persistence or occurrence of leg cramps in different ways, so results could not be pooled. Moderate-certainty evidence from the Thai study suggests that women receiving magnesium are more likely to experience a 50% reduction in the number of leg cramps (1 trial, 86 women; RR: 1.42, 95% CI: 1.09–1.86). The same direction of effect was found in the Swedish study, which reported the outcome “no leg cramps” after treatment, but the evidence was of low certainty (1 trial, 69 women; RR: 5.66, 95% CI: 1.35–23.68). Low-certainty evidence suggests that oral magnesium has little or no effect on the occurrence of potential side-effects, including nausea, diarrhoea, flatulence and bloating. Evidence from the third study was judged to be very uncertain.

Oral calcium versus no treatment

Calcium, 1 g twice daily for two weeks, was compared with no treatment in one small study. Low-certainty evidence suggests that women receiving calcium treatment are more likely to experience no leg cramps after treatment (43 women; RR: 8.59, 95% CI: 1.19–62.07).

Oral calcium versus vitamin C

Low-certainty evidence suggests that there may be little or no difference between calcium and vitamin C in the effect (if any) on complete symptom relief from leg cramps (RR: 1.33, 95% CI: 0.53–3.38).

Oral vitamin B1 and B6 versus no treatment

One study evaluated this comparison, with 21 women receiving vitamin B1 (100 mg) plus B6 (40 mg) once daily for two weeks and 21 women receiving no treatment; however, the low-certainty findings are contradictory and difficult to interpret.

Additional considerations

• The review found no evidence on non-pharmacological therapies, such as muscle stretching, massage, relaxation, heat therapy and dorsiflexion of the foot.

Any exercise (plus standard care) versus standard care

Seven trials (645 women) contributed data to this comparison for low back pain. Trials were conducted in Brazil, the Islamic Republic of Iran, Norway, South Africa and Thailand. Exercise interventions varied from individually supervised exercise to group exercise, including yoga and aqua-aerobics, and some included education via CDs and booklets. Interventions ran for 8–12 weeks and the presence or intensity of pain was assessed in most trials using visual analogue scales. However, the evidence on symptom relief from a meta-analysis of these seven studies is very uncertain. Low-certainty evidence suggests that functional disability scores are better with exercise interventions for low back pain (2 trials, 146 women; standardized MD: 0.56 lower, 95% CI: 0.23–0.89 lower). Evidence on pain intensity (symptom scores) for low back pain was assessed as very uncertain.

Low-certainty evidence suggests that an 8- to 12-week exercise programme may reduce low back and pelvic pain compared with standard care (4 trials, 1176 women; RR: 0.66, 95% CI: 0.45–0.97) and moderate-certainty evidence shows that healthy pregnant women taking part in an exercise programme are probably less likely to take sick leave related to low back and pelvic pain (2 trials, 1062 women; RR: 0.76; 95% CI: 0.62–0.94).

Acupuncture (plus standard care) versus sham acupuncture (plus standard care)

Four small studies conducted in Sweden and the USA evaluated the effects of acupuncture plus standard care versus sham acupuncture plus standard care. However, little data were extracted from these studies and data could not be pooled. Low-certainty evidence from one study suggests that acupuncture may relieve low back and pelvic pain (72 women; RR: 4.16, 95% CI: 1.77–9.78). Evidence from other studies was variously reported and very uncertain.

Acupuncture (plus standard care) versus individualized physiotherapy (plus standard care)

One small study conducted in Sweden involving 46 women with low back and pelvic pain evaluated this comparison. Women’s satisfaction with treatment was the main outcome, but the evidence was assessed as very uncertain.

Osteopathic manipulation therapy (OMT) (plus standard care) versus no osteopathic manipulation (standard care)

Three studies evaluated OMT; however, data could not be pooled and the evidence from individual studies is inconsistent. The largest study involving 400 women compared OMT plus standard care with placebo ultrasound plus standard care, or standard care only. Limited data from this study suggests that OMT may relieve low back pain symptoms more than standard care, and may lead to lower functional disability scores, but may not be better than placebo ultrasound for these outcomes.

One type of support belt versus another type

One small study conducted in Australia compared two types of support belts in womewith low back pain, the BellyBra^® and Tubigrip^® (N = 94) and the evidence from this study was assessed as very low-certainty evidence.

Multimodal interventions versus standard care

One study in the USA reported the effect of a multimodal intervention that included weekly manual therapy by a chiropractic specialist, combined with daily exercise at home, and education versus standard care (rest, exercise, heat pads and analgesics) on low back and pelvic pain. Moderate-certainty evidence suggests that the multimodal intervention is probably associated with better pain scores (1 study, 169 women; MD: 2.70 lower, 95% CI: 1.86–3.54 lower) and better functional disability scores (MD: 1.40 lower; 95% CI: 0.71–2.09 lower) compared with standard care.

Additional considerations

• It is not clear whether the evidence on exercise interventions applies equally to low back pain and pelvic pain, or equally to prevention and treatment, as data from studies of prevention and treatment were pooled. Evidence from two studies on the effect of exercise plus education suggests that such interventions may have little or no effect on preventing pelvic pain (RR: 0.97; 95% CI: 0.77–1.23).
• Very low-certainty evidence on a number of other interventions, such as transcutaneous electrical nerve stimulation (TENS), progressive muscle relaxation with music, craniosacral therapy, and acetaminophen (paracetamol) – which were evaluated in single small trials with apparent relief of symptoms relative to standard care – was also presented in the review.
• Standard care of low back and pelvic pain symptoms usually comprises rest, hot or cold compresses, and paracetamol analgesia.
• There is a paucity of evidence on potential side-effects of alternative therapies, e.g. chiropractic and osteopathic manipulation, and further high-quality research is needed to establish whether these therapies are beneficial for low back and/or pelvic pain and safe during pregnancy.
• Exercise in pregnancy has been shown to have other benefits for pregnant women, including reducing excessive gestational weight gain (see Recommendation A.9).

Fibre supplementation versus no intervention

Evidence from the small study evaluating fibre supplementation versus no intervention on constipation relief (reported as mean frequency of stools) was assessed as being very uncertain.

Stimulant laxatives versus bulk-forming laxatives

Two stimulant laxatives were used in this 1970s study, senna and Normax^®. The latter (containing dantron) is potentially carcinogenic and now only used in terminally ill people; however, data on stimulant laxatives were not available separately for senna. Evidence on relative symptom relief, side-effects (abdominal discomfort, diarrhoea), and maternal satisfaction for stimulant laxatives versus bulk-forming laxatives (sterculia with or without frangula) was assessed as being very uncertain.

Additional considerations

• Various bulk-forming (wheat bran or oat bran fibre supplements, sterculia, methylcellulose, ispaghula husk), osmotic (lactulose) and stimulant laxatives (senna) are available as over-the-counter medications for constipation and are not known to be harmful in pregnancy (166).
• The absorption of vitamins and mineral supplements could potentially be compromised by laxatives.

Pharmacological interventions versus placebo or no intervention

Only one small trial conducted in 1975 (69 women) contributed data. Low-certainty evidence from this trial suggests that rutoside may reduce symptoms (nocturnal cramps, paraesthesia, tiredness) associated with varicose veins compared with placebo (69 women; RR: 1.89, 95% CI: 1.11–3.22). However, no side-effect data were reported.

Non-pharmacological interventions versus placebo or no intervention

Low-certainty evidence suggests that reflexology may reduce oedema symptoms compared with rest only (55 women; RR: 9.09, 95% CI: 1.41–58.54) and that water immersion may reduce oedema symptoms (leg volume) compared with leg elevation (32 women; RR: 0.43, 95% CI: 0.22–0.83). Low-certainty evidence suggests that there may be little or no difference in oedema symptoms (measured as lower leg circumference in centimetres) between foot massage and no intervention (80 women; MD in cm: 0.11 less, 95% CI: 1.02 less to 0.80 more) and between intermittent pneumatic compression and rest (measured as mean leg volume, unit of analysis unclear) (35 women; MD: 258.8 lower, 95% CI: 566.91 lower to 49.31 higher). Only one study (reflexology versus rest) evaluated women’s satisfaction, but the evidence is of very low certainty.

Additional considerations

• Compression stockings combined with leg elevation is the most common non-surgical management for varicose veins and oedema; however, the Cochrane review found no evidence on this practice in pregnancy (160). Compression stockings are also widely used to prevent morbidity in non-pregnant people with varicose veins and the evidence for this practice in a related Cochrane review of compression stockings was generally very uncertain (167).

Maternal outcomes

With regard to maternal satisfaction, moderate-certainty evidence indicates that women who carry their own case notes are probably more likely to feel in control of their pregnancy experience than women whose records are facility-held (2 trials, 450 women; RR: 1.56, 95% CI: 1.18–2.06). Low-certainty evidence suggests that women-held case notes may have little or no effect on women’s satisfaction with ANC (2 trials, 698 women; RR: 1.09, 95% CI: 0.92–1.29). Evidence on caesarean section was very uncertain and other guideline outcomes were not reported in the review.

Fetal and neonatal outcomes

Low-certainty evidence suggests that women-held case notes may have little or no effect on perinatal mortality (2 trials, 713 women; RR: 0.77, 95% CI: 0.17–3.48). No other fetal and neonatal outcomes were reported in the review.

Coverage outcomes

Low-certainty evidence suggests that women-held case notes may have little or no effect on ANC coverage of four or more visits (1 trial, 501 women; RR: 1.25, 95% CI: 0.31–5.00).

Additional considerations

• Other evidence from the review suggests that there may be little or no difference in the risk of case notes being lost or left at home for a visit (2 trials, 347 women; RR: 0.38, 95% CI: 0.04–3.84).
• A WHO multicentre cohort study of home-based maternal records (HBMR), involving 590 862 women in Egypt, India, Pakistan, Philippines, Senegal, Sri Lanka, Yemen and Zambia, was conducted between 1984 and 1988 (176). The study reported that “The introduction of the HBMR increased the diagnosis and referral of at-risk pregnant women and newborn infants, improved family planning and health education, increased tetanus toxoid immunization, and provided a means of collecting health information in the community. The HBMR was liked by mothers, community health workers and other health-care personnel because, by using it, the mothers became more involved in looking after their own health and that of their babies.”

Maternal outcomes

Moderate-certainty evidence shows that MLCC compared with other models of care probably slightly increases the chance of a vaginal birth (12 trials, 16 687 participants; RR: 1.05, 95% CI: 1.03–1.07).

MLCC may reduce caesarean sections (14 trials, 17 674 participants; RR: 0.92, 95% CI: 0.84–1.00), however, this evidence is of low certainty and includes the possibility of no effect. Low-certainty evidence suggests that MLCC models may be associated with lower rates of instrumental vaginal delivery than other models (13 trials, 17 501 participants; RR: 0.90, 95% CI: 0.83–0.97).

Maternal satisfaction: The Cochrane review tabulated data on women’s satisfaction pertaining to various aspects of antenatal, intrapartum and postnatal care. A meta-analysis on satisfaction with ANC only was performed for the purposes of this guideline (see EB Table E.2), the findings of which suggest that MLCC models may increase the proportion of women reporting high levels of satisfaction with the ANC compared with other models (4 trials, 5419 women; RR: 1.31, 95% CI: 1.11–1.54; low-certainty evidence).

Fetal and neonatal outcomes

Moderate-certainty evidence indicates that MLCC probably reduces the risk of preterm birth (8 trials, 13 338 participants; RR: 0.76, 95% CI: 0.64–0.91) and probably reduces perinatal mortality (defined in the review as fetal loss after 24 weeks of gestation and neonatal death) (13 trials, 17 561 women; RR: 0.84, 95% CI: 0.71–0.99). However, low-certainty evidence suggests that it may have little or no effect on low birth weight (7 trials, 11 458 women; RR: 0.96, 95% CI: 0.82–1.13). Evidence on other ANC guideline outcomes was not available in the review.

Additional considerations

• Although the mechanism for the probable reduction in preterm birth and perinatal death is unclear, the GDG considered the consistency of the results and the absence of harm to be important.

Maternal outcomes

Moderate-certainty evidence indicates that group ANC probably does not have an important effect on vaginal birth rates compared with individual ANC (1 trial, 322 women; RR: 0.96, 95% CI: 0.80–1.15). But low-certainty evidence suggests that it may lead to higher women’s satisfaction scores (1 trial, 993 women; MD: 4.9, 95% CI: 3.10–6.70).

Fetal and neonatal outcomes

Moderate-certainty evidence indicates that group ANC probably has little or no effect on low birth weight (3 trials, 1935 neonates; RR: 0.92, 95% CI: 0.68–1.23) and low-certainty evidence suggests that it may have little or no effect on perinatal mortality (3 trials, 1943 neonates; RR: 0.63, 95% CI: 0.32–1.25). However, low-certainty evidence also suggests that group ANC may reduce preterm birth (3 trials, 1888 women; RR: 0.75, 95% CI: 0.57–1.00); this evidence includes the possibility of no effect. Evidence on the risk of having an SGA neonate is of a very low certainty.

Additional considerations

• There is little evidence on the effects of group ANC from LMICs. However, a feasibility study conducted in Ghana suggests that group ANC might improve women’s pregnancy experiences, and providers’ experiences, and potentially improve health outcomes in low-income settings, due to improved health literacy and better engagement of pregnant women with ANC (179).
• It is plausible that group ANC may have an impact on other outcomes outside the scope of the ANC guideline, such as breastfeeding initiation and postnatal contraception, by improving communication and social support related to these healthy behaviours; but the evidence on these potential effects is limited (180).

Effects of community mobilization through facilitated PLA cycles and women’s groups versus standard care (EB Table E.4.1)

The evidence on the effects of community mobilization interventions was synthesized for this guideline from data derived from a Cochrane review of health system and community-level interventions for improving ANC coverage and health outcomes (175). Seven cluster-RCTs conducted between 1999 and 2011, involving approximately 116 805 women, contributed data to this comparison. Trials were conducted in Bangladesh (2), India (2), Malawi (2) and Nepal (1), and six out of seven were conducted in low-resource, rural settings (184–190). The intervention consisted of involving women (pregnant and non-pregnant) in PLA cycles facilitated by trained facilitators, with the aim of identifying, prioritizing and addressing problems women face around pregnancy, childbirth and after birth, and empowering women to seek care and choose healthy pregnancy and newborn care behaviours (191). Meetings were usually held on a monthly basis and specific activities were prioritized according to the local context and conditions. Coverage of women’s group meetings ranged from one group per 309 to one group per 1414 people in the population among included trials, with the proportion of pregnant women attending groups ranging from 2% to 51%. Five out of seven trials were conducted against a backdrop of context-specific health system strengthening in both intervention and control arms; these included training of TBAs and provision of basic equipment to TBAs and/or primary care facilities in four trials. Random effects models were used and sensitivity analyses were performed by including only those trials in which pregnant women comprised more than 30% of the women’s groups.

Values

See “Women’s values” at the beginning of section 3.E: Background (p. 86).

Resources, Equity, Acceptability and Feasibility

See the “Summary of evidence and considerations” for Recommendation E.4.2.

Effects of communication and support provided to women through community mobilization and home visits during pregnancy versus standard care (EB Table E.4.2)

The evidence on the effects of community mobilization and antenatal home visits was synthesized from data derived from a Cochrane review of health system and community-level interventions for improving ANC coverage and health outcomes (175). Four large cluster-RCTs conducted in rural Bangladesh, India and Pakistan contributed data on packages of interventions involving community mobilization and antenatal home visits versus no intervention (192–195). Health system strengthening occurred in both the intervention and control groups in two of the trials. The focus of these packages was generally to promote maternal health education, ANC attendance and other care-seeking behaviour, tetanus toxoid vaccinations and iron and folic acid supplements, and birth and newborn-care preparedness. Household visits were performed by trained lay health workers and consisted of at least two visits during pregnancy. In two trials, these visits were targeted to occur at 12–16 weeks of gestation and 32–34 weeks; in one trial, these visits both occurred in the third trimester; and in the fourth trial the timing of the visits was not specified. Multilevel community mobilization strategies included advocacy work with community stakeholders (community leaders, teachers, and other respected members), TBAs, husbands or partners, and households (husbands or partners, women, and other family members). Two intervention packages included group education sessions for women focusing on key knowledge and behaviour around pregnancy and early neonatal care, including promotion of ANC and other health education. One intervention package included husband education via booklets and audio cassettes. Training of TBAs to recognize common obstetric and newborn emergencies was a component of three intervention packages. In one trial, telecommunication systems with transport linkages were also set up as part of the intervention package. In another trial, community health committees were encouraged to establish an emergency transport fund and use local vehicles, in addition to advocacy work, household visits and women’s meetings.

Additional considerations

• The GDG also considered evidence on antenatal home visits as a stand-alone intervention, but did not make a separate recommendation on this intervention due to the lack of evidence of benefits related to the ANC guideline outcomes. In brief, evidence of moderate- to high-certainty suggests that stand-alone antenatal home visits have little or no effect on ANC visit coverage of at least four visits (4 trials; RR: 1.09, 95% CI: 0.99–1.22), facility-based birth (4 trials; RR: 1.08, 95% CI: 0.87–1.35), perinatal mortality (4 trials; RR: 0.91, 95% CI: 0.79–1.05) and preterm birth (1 trial; RR: 0.88, 95% CI: 0.54–1.44) (see Web supplement).
• The 2013 WHO recommendations on postnatal care of the mother and newborn include the following recommendation:

  “Home visits in the first week after birth are recommended for care of the mother and newborn (strong recommendation based on high-quality evidence for newborns and low-quality evidence for mothers).” This recommendation is accompanied by the remark “Depending on the existing health system in different settings, these home visits can be made by midwives, other skilled providers or well trained and supervised CHWs [community health workers]” (196).

• The 2011 WHO guidelines on Preventing early pregnancy and poor reproductive outcomes among adolescents in developing countries strongly recommend the following in relation to the outcome “Increase use of skilled antenatal, childbirth and postnatal care among adolescents”:

  –

  “Provide information to all pregnant adolescents and other stakeholders about the importance of utilizing skilled antenatal care.”

  –

  “Provide information to all pregnant adolescents and other stakeholders about the importance of utilizing skilled childbirth care.”

  –

  “Promote birth and emergency preparedness in antenatal care strategies for pregnant adolescents (in household, community and health facility settings)” (197).

• Several WHO recommendations included in the 2015 WHO recommendations on health promotion interventions for maternal and newborn health are relevant to community-based interventions to improve communication and support for women during pregnancy (198) – these are presented in Box 3.

Box 3

Relevant recommendations from the 2015 WHO recommendations on health promotion interventions for maternal and newborn health.

Values

See “Women’s values” at the beginning of section 3.E: Background (p. 86).

Resources

A systematic review of the cost–effectiveness of strategies to improve the utilization and provision of maternal and newborn health care in low- and lower-middle-income countries reported that there was reasonably strong evidence for the cost–effectiveness of the use of PLA cycles (199). Estimated costs per life saved for PLA cycle interventions alone was US$ 268 and for community mobilization combined with home visits during pregnancy and/or health system strengthening, costs ranged from US$ 707 to US$ 1489 per death averted. However, costs of these interventions are difficult to estimate and depend on context. Costing must also take into account the facilitators’ time, training and supervision; these elements are considered key to the quality of implementation and the success of the intervention.

Equity

Interventions such as PLA cycles, community mobilization and home visits during pregnancy are a way of facilitating dialogue and action with, and empowering, disadvantaged populations to engage in efforts to improve health and to strengthen broader community support. The women’s groups PLA cycles, in particular, were conducted in marginalized areas where other support mechanisms often do not exist. Interventions to engage male partners/husbands and others in the community to support women to make healthy choices for themselves and their children may help to address inequalities. However, when engaging men, it is important to consider women’s preferences, as including male partners could also have a negative effect for women who would prefer to discuss pregnancy-related and other matters without their partner’s involvement.

Acceptability

Qualitative evidence suggests that women in a variety of settings and contexts readily engage with interventions designed to increase communication and support, provided they are delivered in a caring and respectful manner (high confidence in the evidence) (22). The use of women’s groups is likely to fulfil two key requirements of ANC from a woman’s perspective – the opportunity to receive and share relevant information and the opportunity to develop supportive relationships with other women and health-care providers (high confidence in the evidence). Evidence from women and providers in LMICs also highlighted the importance of active community engagement in the design and delivery of informational-based services, especially in communities where traditional beliefs may differ from conventional understandings (moderate confidence in the evidence). Qualitative evidence from providers suggests that there is a willingness to supply pregnancy-related information and offer psychological/emotional support to women provided that resources are available (high confidence in the evidence) and the services are delivered in a coordinated, organized manner with appropriate managerial support (moderate confidence in the evidence) (45).

Feasibility

Qualitative evidence suggests that, where health-care providers are involved in facilitating women’s groups, they may need additional training to help with the facilitative components and this may be a barrier in some resource-poor settings (high confidence in the evidence). Similarly, the extra costs associated with home visits in terms of additional staff and extra resources may limit implementation in some LMICs (high confidence in the evidence) (45). It has been suggested that community-based interventions introduced through existing public sector health workers and local health systems may be more feasible and more likely to succeed than project-based interventions (200).

Maternal outcomes

High-certainty evidence shows that FANC had little or no effect on caesarean section rates (1 trial, 24 526 women; RR: 1.00, 95% CI: 0.89–1.11), and low-certainty evidence suggests that it may make little or no difference to maternal mortality (3 trials, 51 504 women; RR: 1.13, 95% CI: 0.5–2.57).

With regard to maternal satisfaction, outcomes were reported narratively in the review, as data were sparse. In a survey conducted among a subset of women participating in the WHO trial, fewer women were satisfied with the frequency of visits in the FANC model than in the standard model (77.4% versus 87.2%) and women in the FANC model were less likely to be satisfied with the spacing between visits compared with the standard model (72.7% versus 81%). This evidence was not formally graded due to insufficient data.

Fetal and neonatal outcomes

Moderate-certainty evidence indicates that FANC probably increases perinatal mortality compared with “standard” ANC with more visits (3 trials; RR: 1.15, 95% CI: 1.01–1.32). Based on this RR, the illustrative impact on perinatal mortality rates are shown in Box 4.

Box 4

Illustration of the impact of focused ANC (FANC) on perinatal mortality rates (PMR).

Moderate-certainty evidence indicates that FANC probably has little or no effect on preterm birth (3 trials; RR: 0.99, 95% CI: 0.91–1.08) and low birth weight (3 trials; RR: 1.04, 95% CI: 0.97–1.12) compared with “standard” ANC. In addition, low-certainty evidence suggests that FANC probably makes little or no difference to SGA (3 trials; RR: 1.01, 95% CI: 0.88–1.17).

Additional considerations

• The GDG noted that the review authors explored reasons for the effect on perinatal mortality and the effect persisted in various exploratory analyses.
• In 2012, the WHO undertook a secondary analysis of perinatal mortality data from the WHO FANC trial (205). This secondary analysis, which included 18 365 low-risk and 6160 high-risk women, found an increase in the overall risk of perinatal mortality between 32 and 36 weeks of gestation with FANC compared with “standard” ANC in both low- and high-risk populations.
• It is not clear whether the philosophy of the FANC approach, with regard to improving quality of care at each ANC visit, was implemented effectively in the trials. However, if this element is neglected, a poorly executed FANC model may then simply represent reduced health provider contact, and a reduced opportunity to detect risk factors and complications, and to address women’s concerns.
• The GDG panel considered unpublished findings of a two-year audit of perinatal mortality from the Mpumalanga region of South Africa that has implemented the FANC model (206). The audit from September 2013 to August 2015 comprised data of 149 308 births of neonates weighing more than 1000 g, among which there were 3893 perinatal deaths (giving a PMR of 24.8 per 1000 births). Stillbirth risk was plotted according to gestational age and three peaks in the occurrence of stillbirths were noted, one at around 31 weeks of gestation, another at around 37 weeks, and the third occurring at 40 weeks or more. When these data were compared with stillbirth data from another South Africa province, which uses a model of ANC that includes fortnightly ANC visits from 28 weeks of gestation, the latter showed a gradual rise in the overall stillbirth risk from 28 weeks, with a single (and lower) peak at 40 weeks or more, i.e. no additional peaks at 30 and 37 weeks. These data are consistent with those from the secondary analysis of the WHO trial and suggest that additional visits in the third trimester may prevent stillbirths.
• The GDG also considered the evidence from the Cochrane review on reduced visit ANC models of at least eight visits versus “standard” ANC models with 11–15 visits from four RCTs in HICs (203). Low-certainty evidence suggested that the reduced-visit model (with at least eight visits) may be associated with increased preterm birth (3 trials; RR: 1.24, 1.01–1.52), but no other important effects on health outcomes were noted. In general, however, evidence from these individual studies also suggests that the reduced-visit models may be associated with lower women’s satisfaction.
• The GDG considered unpublished evidence from four country case studies (Argentina, Kenya, Thailand and the United Republic of Tanzania) where the FANC model has been implemented (207). Provider compliance was noted to be problematic in some settings, as were shortages of equipment, supplies and staff. Integration of services was found to be particularly challenging, especially in settings with a high prevalence of endemic infections (e.g. malaria, TB, sexually transmitted infections, helminthiasis). Guidance on implementation of the FANC model in such settings was found to be inadequate, as was the amount of time allowed within the four-visit model to provide integrated care.
• Findings on provider compliance from these case studies are consistent with published findings from rural Burkina Faso, Uganda and the United Republic of Tanzania (208). Health-care providers in this study were found to variably omit certain practices from the FANC model, including blood pressure measurement and provision of information on danger signs, and to spend less than 15 minutes per ANC visit. Such reports suggest that fitting all the components of the FANC model into four visits is difficult to achieve in some low-resource settings where services are already overstretched. In addition, in low-resource settings, when the target is set at four ANC visits, due to the various barriers to ANC use, far fewer than four visits may actually be achieved.
• Programmatic evidence from Ghana and Kenya indicates similar levels of satisfaction between FANC and standard ANC, with sources of dissatisfaction with both models being long waiting times and costs associated with care (209, 210).
• Emotional and psychosocial needs are variable and the needs of vulnerable groups (including adolescent girls, displaced and war-affected women, women with disabilities, women with mental health concerns, women living with HIV, sex workers, ethnic and racial minorities, among others) can be greater than for other women. Therefore, the number and content of visits should be adaptable to local context and to the individual woman.