Global report on preterm birth and stillbirth (3 of 7): evidence for effectiveness of interventions

A meta-analysis of observational studies [9] showed a 40% increase in preterm births when the birth interval was shorter than 6 months, relative to babies born after a birth interval of 18-23 months (RR 1.40; CI 1.24-1.58). The corresponding odds ratios were 1.61 (95% CI 1.391.86) for low birth weight and 1.26 (95% CI 1.18-1.33) for IUGR. A 20% increase in preterm births was also found when the birth interval was 60 months or longer (OR 1.20; 95% CI 1.17-1.24). There are no intervention studies, however, showing that promoting birth spacing will reduce either preterm births or low birth weight.

Stephansson et al. [13], using a logistic regression analysis on the Swedish registry data, showed that compared with interpregnancy intervals (IPIs) 12-25 months, very short IPIs (0-3 months) were linked to an increased risk of stillbirth, though nonsignificant (adjusted OR = 1.3; 95% CI: 0.8-2.1). IPIs greater than 72 months also increased women's risk of stillbirth in this study (adjusted OR = 1.5; 95% CI: 1.1-2.1). Other studies have also shown a link between increased risk of fetal loss and short or long IPIs [14, 15].

The quality of evidence relating birth spacing to preterm birth and low birth weight is moderate, as it is based on a large number of carefully conducted observational studies. While it is not expected that randomized trials will be conducted on the value of birth spacing in the future, there is a critical need for evaluating the effectiveness of various strategies to promote birth spacing, induced terminations and prevention of unwanted pregnancies [16]. The level of recommendation is, therefore, weak in favor of this intervention to prevent preterm birth and low birth weight.

The overall quality of evidence of the impact of birth spacing on perinatal mortality is moderate, but data on stillbirths are sparse. Therefore, this intervention has low evidence of benefit for preventing stillbirth and the recommendation is weak in favor of its use for this outcome. Further evaluation of interventions promoting birth spacing is needed with reference to stillbirth outcomes. However, birth spacing is strongly recommended due to its important effects on both maternal and child health outcomes [12].

No intervention studies show an effect of periconceptional folate supplementation on preterm birth or low birth weight. Three observational studies conducted in the United States were identified in this review. The first found a two-fold increased risk of preterm birth and low birth weight in women with low daily folate intake and low serum folate levels at 28 weeks of gestation [19]. Another observational study, conducted before and after folate fortification programs in California, found small reductions in adjusted risks after the fortification for low birth weight (RR 0.94; 0.93-0.96) and preterm birth (RR 0.96; 0.94-0.97) [20]. A third observational study [21] showed that preconceptional folate supplementation for longer than one year was associated with a reduced incidence of spontaneous preterm birth between 28 and 32 weeks (hazard ratio 0.45, 0.23-0.85), but no significant effect was noted between 32-36 weeks of gestation.

A Cochrane review, including three trials that reported perinatal outcomes, concluded that periconceptional folic acid supplementation is of proven benefit in reducing neural tube defects, reflected by a statistically significant reduction in its rate (RR 0.28; 95% CI 0.13-0.58), but impact on stillbirths is that of a 22%, nonsignificant reduction (RR 0.78; 95% CI 0.34-1.78) [22].

The quality of evidence is low for preterm birth, but because the three observational studies point in the same direction, the recommendation is weak in favor of the intervention. The quality of evidence is moderate for stillbirths given the wide uncertainty in the results; the intervention cannot be recommended exclusively for preventi ng stillbirth and thus received a weak recommendation against the intervention for this outcome. Nevertheless, periconceptional folate is strongly recommended for other MNCH outcomes due to its protective effect on neural tube defects. Further studies are needed to assess potential long-term benefits of folic acid supplementation during pregnancy.

Observational studies in Guatemala, Pakistan and Zimbabwe show an association between this practice and decreased birth weight, but no information is available on preterm births [24, 25]. In addition, no intervention studies are available on the impact of reduced indoor air pollution on either low birth weight or preterm birth.

Studies on the impact of indoor air pollution on stillbirths are restricted to South Asian countries. Mavalankar, in India, found a nonsignificant increase in stillbirth risk in a case control study among women exposed to smoke (OR 1.5; 95% CI 1.0-2.1) [26]. Another population-based Indian study showed that women using biomass fuels had a significantly higher risk of stillbirth than those utilizing cleaner fuels (OR 1.44; 95% CI 1.04-1.97) [27]. Siddiqui et al. from Pakistan reported nearly a two-fold greater risk of stillbirths in pregnant women exposed to biomass fuel (OR 1.90; 95% CI 1.10-3.20) [28].

There is a lack of studies relating indoor air pollution to preterm birth, especially randomized controlled trials [29, 30]. Therefore the quality of evidence is very low and the recommendation is weak against the intervention for this particular outcome. For low birth weight and stillbirths, the quality of the evidence is low, as it is largely based on observational studies. However, the intervention is strongly recommended for the prevention of respiratory infections [31, 32]. Future studies of measures to reduce indoor air pollution must also measure pregnancy and neonatal outcomes.

Cigarette smoking is a well-known cause of preterm birth and intrauterine growth restriction [40]. In a Cochrane review published in 2004, smoking cessation interventions significantly reduced low birth weight (RR 0.81; 0.70-0.94) and preterm birth (RR 0.84; 0.72-0.98), as well as smoking during pregnancy [41]. It should be noted, however, that the experience of smoking cessation programs is predominantly from high-income countries [42].

Notwithstanding the established benefits to mother and fetus, there are few studies that have reported the effect of smoking cessation on perinatal outcomes and stillbirths. One cohort study from Sweden shows increased risk of stillbirths among smokers [43].

The quality of evidence is high for both preterm birth and low birth weight, and the intervention has a strong recommendation, especially in countries where smoking has a high prevalence during pregnancy. The strong recommendation is reinforced by other well-known effects of smoking on health. The lack of studies on smoking cessation and stillbirth outcomes points to a data gap that should be filled. Maternal exposure to second-hand smoke is an additional research gap. It should also be noted that all studies reviewed originate from high-income settings, and therefore intervention studies in low- and middle-income settings are needed.

A Cochrane review of seven trials of balanced protein energy supplementation for pregnant women did not show a significant effect on preterm births (RR 0.83; 0.65-1.06), but intrauterine growth restriction was reduced by 32% (RR 0.68; 0.54-0.86) [45]. Five of the seven trials were carried out in LMICs.

The same review showed a significant reduction in the risks of stillbirth (RR 0.55; 95% CI 0.31-0.97) and a near-significant reduction in neonatal deaths (RR 0.62; 95% CI 0.37-1.05) [45].

The quality of evidence is high and the intervention is not recommended for preventing preterm birth. However, this intervention is strongly recommended in appropriate populations (food insecure households and mothers with low body mass index) due to its affect on fetal growth and on preventing stillbirth.

A review of nine RCTs comparing multiple micronutrient supplementation during pregnancy with either no supplements, two or fewer micronutrients, or placebo, showed a reduction in low birth weight (RR 0.84; 95% CI 0.74-0.95), IUGR (RR 0.92; 95% CI 0.86-0.99) and maternal anemia (RR 0.61; 0.52-0.71), but no effect on preterm birth [50]. A more recent trial in Tanzania compared multivitamins with placebo; the intervention group showed a reduction in low birth weight (RR 0.82; 95% CI 0.70-0.95) and in the frequency of small for gestational age newborns (RR 0.77; 95% CI 0.68-0.87), but no effect on preterm births or fetal mortality [51].

A 2006 meta-analyses of multiple micronutrient supplementation compared with iron and folic acid found no differences in maternal or neonatal outcomes [50]. However, when a 2008 RCT from Indonesia [52] was added to this meta-analysis, a significant reduction in low birth weight was observed (OR 0.84; 95% CI 0.74-0.95) [53].

A reasonably large number of studies have recently evaluated the impact of multiple micronutrients on perinatal mortality and stillbirth outcomes. A recent meta-analysis was conducted, including trials published since the most recent Cochrane reviews, of the impact of multiple micronutrient supplementation in pregnancy [54]. The meta-analysis (nine RCTs, N=40,222 women, N=20,277 intervention group, N=19,945 controls) compared the impact on stillbirths of multiple micronutrient supplementation during pregnancy (intervention) with either iron or iron and folate (controls) and found a nonsignificant trend toward reduced stillbirths among the intervention group versus the control group (RR 0.91, 95% CI: 0.80-1.03).

One note of caution was provided by the report of increased birth asphyxia among babies in the intervention group in two micronutrient supplementation RCTs carried out in Nepal [55‐57]. Their pooled analysis showed increased rates of perinatal and neonatal mortality (RR 1.52; 95% CI 1.03-2.25) in the intervention group. However, such an effect was not found in the Indonesia randomized trial, where infant mortality was 18% lower among infants whose mothers were supplemented with micronutrients [52]. In sites where these RCTs were carried out, most births took place either at home or in health posts, with limited access to emergency obstetric care. In Indonesia however, the study area had trained midwives and a signifcnatly greater proportion of births were assisted by skilled birth attendants. Thus this intervention must be viewed in the context of health system functionality and might provide benefits in circumstances where skilled care and facility births are available [58].

There is high quality evidence the intervention has a significant impact on low birth weight but no effect on preterm births or stillbirths. However, due to moderate-quality evidence of increased neonatal mortality in South Asia, more information is necessary from effectiveness trials in different health systems settings before the intervention can be recommended for scaling up.

A meta-analysis of four trials of iron supplementation during pregnancy, with and without folic acid, [60] showed a reduction in anemia but no significant effect on preterm birth (RR 0.76; 95% CI 0.47-1.24) or low birth weight (RR 0.59; 95% CI 0.23-1.49).

The impact of iron or iron-folate supplementation on the prevention of stillbirth is mixed, [54] partly because most studies are underpowered to detect differences. The Cochrane review by Pena-Rosas et al. on antenatal iron supplementation found a nonsignificant risk of increased perinatal mortality with iron supplementation compared to placebo (RR 2.50, 95% CI 0.10-59.88) [60]. Another study from The Gambia, reported stillbirth rates of 2.9% vs. 4.3% in intervention (200 mg oral ferrous sulphate) and control groups, respectively [61].

There is moderate evidence of a lack of effect for iron-folate supplementation on preterm birth, low birthweight and stillbirth. Although this is based on randomized trials, the quality of the evidence is moderate, because the confidence intervals are very wide. The recommendation—based exclusively on these outcomes—is weak against supplementation. Nevertheless, the well known impact of iron and folate in the prevention of maternal anemia result in a strong recommendation in favor of supplementation.

In a Cochrane review of 13 trials, seven of which were from low-middle income countries, zinc supplementation during pregnancy led to a modest reduction of 14% (95% CI 0.76-0.98) in preterm birth, but had no effect on low birth weight (RR 1.05, 95% CI 0.94-1.17), suggesting this effect may be restricted to large preterm babies [64].

The recent Cochrane review of 9000 pregnancies mentioned above had 7 studies that reported no effects on stillbirth and perinatal mortality outcomes. The single study [55] undertaken in Nepal, a population with low maternal zinc status, reported no effect on stillbirths or other perinatal deaths (RR 1.03, 95% CI 0.71-1.51). A recent review of the subject [65] also did not report any stillbirth or other perinatal outcomes [55].

The quality of evidence for zinc supplementation on stillbirth, preterm birth and low birth weight is high. The intervention is not recommended for the prevention of low birth weight or stillbirths, and has a weak recommendation for preventing preterm births, given its small effect size. No firm conclusion was reached on the effect of maternal zinc supplementation on stillbirth.

A meta-analysis of seven intervention trials with oral magnesium treatment starting before the 25th week of gestation showed a lower frequency of preterm birth (RR 0.73, 95% CI 0.57-0.94), as well as a reduction in low birth weight (RR 0.67, 95% CI 0.46 to 0.96) and improved maternal outcomes (hemorrhage and hospital admissions). However, authors of the Cochrane review noted that the studies included in the review had important shortcomings [67].

The Cochrane review [67] included three RCTs with 1954 recipients that reported stillbirth outcomes and showed no impact (RR 1.0, 95% CI 0.29-3.44).

The quality of evidence is moderate for both preterm birth and low birth weight, and there is a weak recommendation against the intervention until better studies become available. There is fairly consistent evidence that magnesium has no impact on stillbirth, and the recommendation is strong against the intervention for this outcome. A recent Cochrane review shows a reduced risk of cerebral palsy in the offspring of high-risk women who received magnesium sulfate [68] and thus this intervention should be evaluated further with a range of alternative outcomes.

Calcium supplementation trials in low-risk women in populations with low-calcium diets reduced the risk of preeclampsia (RR 0.48; 95% CI 0.33-0.69) but its impact on preterm birth (10 trials, RR 0.81; 95% CI 0.64-1.03) and low birth weight (8 trials, RR 0.84, 95% CI 0.68-1.03) was not quite significant [71]. However, when the analysis was restricted to the four small studies including 568 women at high risk of preeclampsia, there was a significant decrease in preterm birth (RR 0.45, 95% CI 0.24 to 0.83).

In a recent small placebo-controlled RCT in low-risk Indian primigravidae with low dietary calcium, calcium supplementation led to less preeclampsia (OR 0.31; 95% CI 0.15-0.63) and significantly fewer preterm births (OR 0.51; 95% CI 0.28-0.93) [72].

The Cochrane review [71] reported stillbirth and neonatal death before discharge in 10 trials with 15,141 participants, but showed no impact (RR 0.89, 95% CI 0.79-1.09).

The quality of evidence for preterm birth and low birth weight is moderate, and the intervention has a weak recommendation for preventing preterm birth. For stillbirth, the quality of evidence is high suggesting a lack of effect. Therefore, the intervention is not recommended for these particular outcomes. However, it is strongly recommended for preventing preeclampsia in populations with low levels of dietary intake of calcium.

Five randomized trials found that marine oil supplementation to low-risk women did not show any effect on preterm births (RR 0.92, 95% CI 0.79, 1.07), low birth weight, or the risk of preeclampsia [74]. Two RCTs showed LCPUFA supplementation was associated with a significantly lower rate of early preterm birth (<34 weeks of gestation) (RR 0.39, 95 % CI 0.18, 0.84) [75].

No reported outcomes.

The quality of evidence is high showing no effect of the intervention to prevent preterm birth and low birth weight. The recommendation against the intervention is strong for these outcomes given the current state of evidence. Further studies are needed on a possible effect on early preterm birth.

There are surprisingly few studies evaluating the relationship of cardiotocographic monitoring with stillbirth and the overall quality of available evidence is low. In light of the probable link with improved outcomes in developed countries, and the widespread use of low-cost CTG monitoring equipment in clinical practice, the current recommendation is weak. Further studies must be conducted with appropriate design and power to assess this intervention in health systems in developing countries.

There is high-quality evidence suggesting a lack of effect on stillbirth, and the recommendation against this intervention is strong.

During pregnancy, observational studies show associations between syphilis and both preterm birth and low birth weight. In Tanzania, women with high-titer active syphilis had a six-fold greater risk of preterm birth and a three-fold increase in low birth weight compared with seronegative women. It was estimated that syphilis accounts for one fourth of the preterm births in this population [88]. An earlier study from Malawi also found an increase of preterm and low birth weight babies among women with syphilis (OR 3.6; 95% CI 1.6-7.9) [89]. All studies reviewed are from LMICs.

Penicillin effectively reduces the risk of congenital syphilis [84, 90], but there are no intervention studies showing an effect of syphilis screening and treatment on preterm birth.

The effectiveness of antibiotics in curing gestational syphilis and preventing congenital infection was established in the 1940s, before RCTs had been adopted [90]. Observational studies in Swaziland, and Kenya suggest syphilis screening and treatment is associated with reduced perinatal mortality [91] and reduced stillbirths [84, 92]. In the Tanzanian study [84], the proportion of stillbirths was higher among women treated for low-titer active syphilis (4.8%) vs. seronegative women (2.5%) (crude OR = 1.95, 95% CI: 0.96 - 4.0); being similar for women treated for high-titer active syphilis (2.3%) versus seronegative ones (2.5%). For ethical reasons it was not possible to have an untreated group with syphilis, but these results suggest that treatment of high-titer women can reduce the risk. The results for low-titer women are difficult to interpret in the absence of an untreated comparison group.

The quality of evidence for preventing preterm birth, low birth weight and stillbirth is moderate, being based on observational studies and accumulated knowledge. Given the current level of knowledge, it is not expected that RCTs will be conducted as these would be unethical. The recommendation is weak in favor of the intervention for preventing preterm birth or low birth weight. The recommendation is strong in favor of recognizing and treating maternal syphilis to reduce stillbirth and congenital syphilis.

Malaria infection may affect fetal growth and gestational duration through maternal anemia and placental infection [93]. A systematic review of interventions with antimalarials during pregnancy showed that - among women in their first or second pregnancies - treatment reduced anemia, parasitaemia, placental malaria, perinatal deaths and low birth weight (six trials, RR 0.57; 95% CI 0.46-0.72). No effect on preterm births was observed in the only trial assessing this outcome [94].

A Cochrane review on prophylactic antimalarials or IPTp showed no overall effect, but when the analyses were restricted to women in their first or second pregnancy, antimalarials significantly reduced perinatal mortality (RR 0.73; 95% CI 0.53-0.99) and were associated with a smaller, nonsignificant reduction in stillbirths (RR 0.87; 95% CI 0.62-1.21) [94].

Although there is little evidence for preterm birth or stillbirth, the quality of evidence is high for using IPTp to reduce low birth weight and perinatal mortality in malaria- endemic areas. IPTp is thus strongly recommended for women in their first or second pregnancy.

A review of five randomized clinical trials of ITNs during pregnancy—four in Africa and one in Thailand—showed a 33% reduction in low birth weight in the African trials, but no effect in Thailand [97]. In the only trial with information on preterm births, conducted in Kenya [98], no effect was demonstrated.

The same Cochrane review showed a 33% reduction in fetal losses (abortions and stillbirths) in the first to fourth pregnancy (RR 0.67; 95% CI 0.47-0.97), but not in fifth and higher-order pregnancies [97, 99].

The quality of evidence is high for low birth weight and stillbirth, and the intervention is strongly recommended for these outcomes. There is very limited evidence—a single RCT—for preventing preterm birth, and the recommendation is weak (against), given that this trial found no effect. Overall, ITNs are strongly recommended due to their effect on maternal and child mortality and morbidity. There are few studies employing a combination of IPTp and ITN on pregnancy outcomes, including preterm birth and stillbirth.

Observational studies show an association between maternal urinary tract infections and both preterm birth and low birth weight. A meta-analysis of four observational studies indicated a decreased risk of preterm birth (RR 0.51, 95% CI 0.36-0.69) in nonbacteriuric pregnant women in comparison to those with bacteriuria [100].

A 2007 meta-analysis evaluated 14 RCTs comparing the effect of antibiotics versus placebo [101] on different outcomes. Treatment reduced maternal pyelonephritis (RR 0.23; 95% CI 0.13-0.41) and low birth weight (7 studies; RR 0.66; 95% CI 0.49-0.89), but did not significantly reduce preterm birth (3 studies; OR 0.37; 95% CI 0.10-1.36). The three studies addressing preterm birth, however, had a number of methodological problems, such as using different cutoff points (<38 weeks in two studies and <37 weeks in one), having small sample sizes, and different diagnostic and treatment criteria.

No reported outcomes.

The quality of evidence for preterm birth is low, but it is high for low birth weight. The intervention is strongly recommended for preventing low birth weight and reducing maternal morbidity, but has a weak recommendation for preventing preterm births. Further studies are required on preterm birth and stillbirth.

Several systematic reviews of screening and treatment for bacterial vaginosis found no impact on preterm births or low birth weight [103‐108]. Because of this, routine pregnancy screening and treatment of bacterial vaginosis is not recommended [109‐111].

However, RCTs in which antibiotics were given before 20 weeks found a significant reduction in preterm births (RR 0.72; 95% CI 0.55-0.95) [107]—but further trials are needed before a recommendation for early treatment can be issued. It is possible the bacteria that cause the infection may ascend into the uterus before or during early pregnancy, starting an inflammatory response that would not be affected by late treatment [39].

No reported outcomes.

There is high quality evidence the treatment of bacterial vaginosis has a lack of effect for preventing preterm birth or low birth weight. A strong recommendation is made against the use of the intervention. The effect of early treatment of bacterial vaginosis on preterm birth and stillbirth is a research gap that needs to be addressed.

Observational studies, particularly those from Europe, report increased rates of preterm birth in women receiving HAART during pregnancy. This risk was particularly pronounced for PI use that started early in pregnancy or before conception. For example, a multicenter European collaborative study including nearly four thousand mothers reported relative risks of preterm birth of 2.60 (95% CI 1.43-4.75) and 1.82 (95% CI 1.13-2.92), for those exposed to combination therapy with and without a protease inhibitors, respectively, compared to no treatment [112]. An observational study in the USA also showed increased rates of preterm birth among women receiving HAART with protease inhibitors, compared with any other treatment option (OR 1.8; 95% CI 1.1-3.0) [113]. However, a 2007 meta-analysis of 1 retrospective and 13 prospective studies, including the aforementioned, showed that antiretroviral therapy during pregnancy is not associated with an overall increased risk of preterm birth [114]. The use of combination regimens before or early in pregnancy may slightly increase the risk of preterm birth, although confounders such as maternal HIV-stage may contribute to this observation. Observational studies suggest an association between preterm birth and use of HAART in pregnancy, in particular initiating PI-based HAART early in gestation, but inadequately control for stage of maternal HIV disease and other potential confounding variables [115‐118].

Fewer studies report on stillbirth outcomes and antiretroviral therapy during pregnancy. A report combining two completed clinical trials and five ongoing, prospective observational studies from the United States found similar rates of stillbirths between women who did and did not receive antiretroviral therapy during pregnancy [119]. Follow-up analysis from 2 of the included observational studies continued to show either no association between ART and stillbirth [113] or a decreased risk for stillbirth (OR 0.06, 95% CI 0.02-0.18) [120]. Townsend et al., 2007 compared women on HAART with women on mono/dual therapy. In comparison with exposure to mono/dual therapy, exposure to HAART was associated with a nonsignificant increased risk of stillbirth (adj. OR=2.27, 95% CI: 0.965.41; p=0.063) [121].

Finally, there was no significant difference in stillbirth rates between groups of women treated with HAART throughout pregnancy and short-course therapy in the two studies originating from low-income countries [122, 123].

The use of combination antiretroviral therapy has dramatically decreased the rate of maternal-to-child HIV transmission. However, observational data suggest a possible association between ART therapy and increased risk of preterm birth. This may be particularly burdensome in low-income countries, particularly those with less access to neonatal care. Large-scale, randomized, controlled trials are needed to define the risk of HAART therapy and adverse pregnancy outcomes.

One RCT in Peru found no difference in low birth weight between women receiving anti-helminthitic therapy (mebendazole) compared to placebo (8.1% and 8.7%, respectively; p=0.7). Both groups received iron supplements. No information on gestational age was available [124]. A recent Cochrane review of the treatment of soil-related helminths in pregnancy shows no statistically significant impact on preterm birth (RR 0.85, 95% CI 0.38-1.87) or low birth weight (RR 0.94, 95% CI 0.61-1.42) [125].

An observational study from Sri Lanka on the effect of mebendazole therapy during pregnancy on birth outcomes showed that stillbirths and perinatal deaths combined were significantly less common among women who received mebendazole as part of antenatal care (RR 0.55; 95% CI 0.4-0.77) [126]. However, the Peruvian RCT did not show statistically significant differences in rates of miscarriage, stillbirth (8/522 in the mebendazole plus iron group vs. 4/520 in placebo plus iron) or early neonatal death [127], but was underpowered for these outcomes.

The quality of evidence is low, and there is a weak recommendation against using anti-helminthics to prevent stillbirth. Very little information is available on preterm birth. Hookworm treatment, however, is recommended for pregnant women to reduce anemia in high-prevalence populations.

A meta-analysis of observational studies indicate associations between periodontal disease and preterm birth [128], but clinical trials show conflicting results [129]. From the four available RCTs, two showed no effect of treatment on preterm birth or low birth weight [130]. The other two described reductions in preterm birth, and one RCT showed a reduction in low birth weight [131, 132].

No reported outcomes.

Although RCTs on the impact of periodontal treatment on preterm birth or low birth weight are available, the quality of the evidence is moderate, because results so far are inconclusive. Currently, the intervention is not recommended for the sole intent of preventing preterm birth or low birth weight. The intervention may be beneficial for improving maternal oral health, but further research on fetal outcomes is needed.

A meta-analysis of six randomized trials comparing the use of progesterone with placebo in high-risk women, showed a reduction of preterm births in the intervention group (RR 0.65, 95% CI 0.54-0.79) [133], as well as a decreased prevalence of low birth weight (four studies, RR 0.63, 95% CI 0.49-0.81). Five of the six trials are from high-income settings.

The same meta-analysis failed to show a significant impact on perinatal deaths (five studies, RR 0.66; 95% CI 0.37-1.19).

The evidence of progesterone use to prevent recurrence of preterm birth is high and the intervention is strongly recommended. For stillbirth, the evidence is moderate due to the wide confidence interval, and the recommendation is weak against this intervention.

In four RCTs of cervical cerclage for high-risk women, reviewed in a Cochrane publication [134], there was no difference in the occurrence of preterm births between intervention and control women (RR 1.04, 95% CI 0.99-1.10).

Another Cochrane review showed a nonsignificant 20% reduction in perinatal loss, death at or after 24 weeks of gestation and up to the first week of neonatal life. (RR 0.80, 95% CI 0.48-1.36) [135]. A more recent systematic review of seven RCTs also found similar results for pregnancy loss or death before hospital discharge (OR 0.81, 95% CI 0.60-1.10) [136].

The quality of evidence is high for both preterm birth and pregnancy loss. The recommendation is strong against the intervention for preventing preterm delivery, and weak against for preventing stillbirth, due to greater uncertainty in the evidence.

A systematic review of four RCTs comparing the use of vitamin A with placebo in HIV-positive pregnant women did not show significant effects on maternal-to-child HIV transmission (OR 1.06; 95% CI 0.89-1.26) or preterm birth (RR 0.88, 95% CI 0.65-1.19). The intervention significantly increased mean birth weight by 89 g (95% CI 85-95) and resulted in a near-significant reduction in low birth weight (RR 0.83, 95% CI 0.68-1.01) [138].

Only one study reported fetal loss (miscarriage and stillbirth) and indicated a reduction in stillbirth with multi-vitamin supplementation among HIV-infected women in pregnancy (30/512 in supplemented groups versus 49/509 in controls (RR 0.61, 95% CI 0.39-0.94) [137].

The quality of evidence for preterm birth and low birth weight is moderate due to uncertainty in the estimates, whereas the quality is low for stillbirth given that only one study is available. The recommendation is weak against this intervention.

There is high quality evidence that community-based interventions that include birth preparedness can prevent stillbirths. However, the effects of birth preparedness per se cannot be assessed. Community-based interventions are strongly recommended in appropriate settings where the proportion of home deliveries is high.

The quality of evidence reviewed for evaluating the partogram for care during delivery is low, given the controlled circumstances in which these studies have been conducted. The recommendation is weak in favor of this intervention. Further RCTs are required in which stillbirth is a primary outcome.

Only two studies are available on this intervention, both from developed countries, with no significant effects on stillbirth. As a consequence, the quality of the evidence is low and there is a strong recommendation against its adoption.

The quality of evidence in favor of c-sections or instrumental deliveries, compared to no such procedures, is moderate for stillbirths. Fifteen of the 40 studies reviewed on this topic were carried out in LMICs. RCTs for testing these approaches would be unethical. The recommendation is strong in favor of the interventions to prevent stillbirths. Excessively high c-section rates, however, should be avoided, as the frequency of preterm delivery and neonatal mortality both rise at rates of caesarean delivery of between 10% and 20% [158].

The quality of evidence for preventing perinatal mortality is high and the recommendation for this intervention is strong.

The quality of evidence for elective induction for post-term pregnancies is high for perinatal mortality, and the recommendation is strong in favor of the intervention in low-risk pregnant women at 41 weeks or more.

The quality of evidence for perinatal mortality is moderate given the wide uncertainty of the estimates, and the recommendation is weak in favor of induction. However, the recommendation is strong for improving other maternal and infant outcomes.

The quality of evidence for perinatal mortality is moderate, and the recommendation is weak against home delivery.

The quality of evidence for preventing preterm birth, morbidity and mortality is high and the intervention is strongly recommended.

The quality of evidence on morbidity and mortality is high and the intervention is strongly recommended to be scaled up to improve preterm survival.

The quality of evidence is high, and there is a strong recommendation against antibiotics for preterm labor with intact membranes.

On the basis of available evidence, a 30-second delay in cord clamping is a safe intervention for preterm newborns. The quality of the evidence is high and the recommendation is strong in favor of this intervention. All studies are from high-income countries.

Different approaches have been used for oxygen delivery to the neonate (face mask, bag mask, nasal cannulae, CPAP) and these may vary in terms of achieving adequate oxygen saturation and hence dictate the need for further management (intubations, CPR etc). Only one study addressed preterm neonates exclusively [189] while others reported outcomes in all newborns.

Capasso et al [190] compared oxygen delivery on intermittent positive pressure with nasal cannulae versus facial mask in primary resuscitation of the newborn with moderate asphyxia, concluding that while resuscitation with nasal cannulae required fewer intubations and chest compressions, it had comparable survival rates and Apgar scores to normal controls. Palme et al [191] compared the efficiency of five widely used face masks on newborns with respect to the rates of leakage and frequency and ease of cleaning. They reported least leakage with circular silicone rubber mask ('Laerdal') and found it easier to clean, boil and autoclave.

Massawe et al [192] found that mouth-to-mask and bag-to-mask ventilation were comparable in terms of Apgar scores, heart rate and time to first breath. The only RCT on preterm newborns [189] evaluated the impact of nasal CPAP on BPD and the need for intubation, reporting a lower number of intubations in neonates given nasal CPAP on admission to the NICU compared to those who were not given nCPAP.

These results suggest that nasal CPAP is preferable to manual inflations in preterm newborns and may lower the need for chest compressions and intubations, whereas mouth-to-mask ventilation is equivalent to bag-to-mask ventilation for neonates. The American Academy of Pediatrics has initiated a Helping Baby's Breathe program (AAP, personal communication 2009) to enable lower level health workers to undertake standardized resuscitation in primary care settings. Validation studies of the protocol are underway.

The crucial role of neonatal resuscitation in immediate newborn care is well accepted [193]. On the other hand, its use has also been associated with adverse long- and short-term outcomes possibly due to survival of severely affected newborns [194‐196]. Given that RCTs of such interventions are difficult, we reviewed seven observational studies from PubMed and one interventional study from Cochrane clinical trials. Segregated data for term and preterm neonates was available for two studies, whereas other authors did not separate these two groups.

Vakrilova et al [197] discussed the need of delivery room cardio-pulmonary resuscitation (CPR) in VLBW and ELBW neonates and reported that birth weight and gestational age are the most important factors, determining the intensity of delivery room CPR and the prognosis in newborns of < 1500 g. Multiple studies reported the long-term outcomes including language symmetry, attention shift, visual attention, neuropsychological sequela and social and educational adjustments. While language symmetry and attention shift [196] were impaired in the individuals given CPR at birth, there was no difference between the resuscitated and non resuscitated groups in terms of other parameters for long-term outcomes [195, 196, 198, 199]. Dorfsman et al [200] compared two thumb technique of neonatal CPR versus American Heart Association recommended two finger technique. He reported that TT (two thumb) chest compression produced higher MAP, systolic and diastolic blood pressures when compared with TF (two finger) chest compression technique during a clinically relevant duration of prolonged CPR.

Sanchez-Torres et al [194] evaluated the impact of CPR provided in delivery room on survival and short term neurological outcomes in preterm infants. The infants receiving CPR and those infants who did not were comparable in terms of mortality, NEC, IVH and BPD. However the infants undergoing CPR needed surfactant and oxygen more frequently. On the other hand Deulofeut et al [201] reported higher rates of mortality, IVH and periventricular leucomalacia in neonates who were provided CPR at birth.

The optimum oxygen concentration required for newborn resuscitation has been the subject of investigation [202]. While 100% oxygen has been traditionally used for all cyanotic infants [203], reports of increasing incidence of hypoxic ischemic encephalopathy (HIE), as well as other secondary outcomes, have made its use debatable. High oxygen levels were also associated with adverse outcomes like high oxidative stress [204] and increased incidence of leukemia [205].

It has been proposed that room air is as effective as 100% oxygen in overcoming neonatal asphyxia, [188] and several studies and systematic reviews addressed this issue [202, 203, 206]; [188, 204, 207‐210].

Three intervention studies on preterm infants [211‐213], all from high-income settings, reported similar mortality and morbidity rates, including BPD, among those resuscitated with room air or 100% oxygen (p=0.95 for NMR) [211], p=0.71 for NMR [212].

Studies which did not separate term and preterm newborns compared 100% oxygen with room air, concluding either that outcomes are similar in both groups, or that the room air group has advantages in terms of morbidity and mortality [202, 203, 206]. Saugstad et al [214] also reported no differences in neurodevelopmental outcomes between neonates resuscitated with room or 100% oxygen and this finding has been replicated in a recent population based study by Hellstrom-Westas et al.[215].

Many studies confirm that training programs for neonatal resuscitation are vital for survival [193, 216, 217].

Impact on preterm infants was observed in three studies where training programs led to significant reductions in terms of mortality and morbidity [218‐220]. Specialized neonatal resuscitation teams have also played an important role in reducing preterm neonatal morbidity and mortality [221].

Additional studies, mostly observational, document the impact of training programs on all newborns, either term or preterm. There is ample documentation on the impact of such programs on improving the performance and technical abilities of health professionals, and on reducing neonatal morbidity and mortality rates [219, 222‐239]. Similar outcomes were also observed by the employment of specialized neonatal resuscitation teams [221, 240, 241].

Finally, large-scale studies support the effectiveness of resuscitation programs. Sen et al [217] demonstrated a 21% reduction in mortality after a sick newborn care unit was set up in Kolkata that included resuscitation facilities. Enweronu-Laryea et al [242] reported similar effect of improved neonatal care facilities on neonatal survival in Ghana which provided improved obstetric services, referral systems, well developed neonatal units and trained staff.

The quality of evidence is moderate given the small number of studies including preterm infants and the uncertainty in estimates. WHO has recently commissioned three large prospective RCTs in community settings to evaluate the impact of neonatal Vitamin A supplementation. Given current knowledge, there is a weak recommendation against its use in preterm infants, but this may well change as the evidence accumulates.

The quality of the evidence for both antenatal and postnatal vitamin K use is moderate. There is a strong recommendation for postnatal vitamin K, and a weak recommendation in favor of antenatal vitamin K for women in premature labor.

Notwithstanding the interesting findings in these studies, the overall quality of the evidence is low regarding the benefit of zinc supplementation in preterm infants, and there is a weak recommendation against the intervention until further studies with adequate samples and gestational age assessment are available.

While some experts have recommended selenium supplementation of preterm formulas (Klein 2002), the quality of the evidence is moderate due to the small number of studies, and there is a weak recommendation in favor of supplementation. Further studies are urgently needed.

Despite the interesting findings from these studies, the overall evidence for the benefits of chlorhexidine use among preterm infants is moderate and further trials are needed to validate impact before recommending this intervention for widespread use.

Available data on community-based case management of neonatal sepsis and pneumonia in LMICs make a compelling case for providing such care in circumstances where referral may not be possible. While the overall evidence is moderate due to the small number of studies reporting on results among preterm infants, additional evidence is provided by studies including newborns in general, whether term or preterm. There is a strong recommendation in favor of this intervention in appropriate settings.

There is a high level of evidence that hospital-based KMC is an effective method of preventing severe morbidity and mortality, especially in low- and middle-income countries. The intervention is inexpensive and feasible, and therefore strongly recommended for hospital settings.

On the other hand, the evidence on community-based KMC is restricted to a single study with less clear results. In addition, that study failed to reach compliance with the recommended practice, which raises issues about its feasibility for widespread implementation [279]. New research is needed on the best ways to scale up the intervention at a community level and for estimating its impact on morbidity and mortality.

There is a high quality of evidence that breast milk has beneficial effects in the prevention of necrotizing enterocolitis and severe infections in preterm babies. On the other hand, the level of evidence of an association between early breastfeeding and neonatal mortality is only moderate. The intervention is strongly recommended to be scaled-up due to its multiple benefits.

The quality of the evidence supporting thermal control relative to absence of control is based on observational studies carried out in the distant past, and according to the GRADE guidelines is judged as moderate - even though RCTs would be unethical. Several different approaches were shown to provide adequate thermal control in RCTs. Of particular relevance to LMICs are skin-to-skin contact and plastic wraps, both of which are superior to routine thermal care in terms of keeping neonates warm. These provide high quality evidence. Comparisons of skin-to-skin contact relative to routine care suggest a possible impact on mortality, but the confidence interval is wide. The overall recommendation for skin-to-skin contact and plastic wraps is strong.

The quality of evidence of nCPAP for treatment of respiratory failure among preterm newborns to prevent death, bronchopulmonary dysplasia, or both is high and this modality is used routinely in developed countries. In addition, the quality of evidence for the use of nCPAP as an alternative to mechanical ventilation is high, making it particularly appealing in LMIC settings even though only two of the 29 studies reviewed are from such settings. The intervention is strongly recommended, however optimal delivery systems, interfaces and optimal pressures must be elucidated. Further, ensuring adequate support services such as managing airleak syndrome will be important prior to widespread implementation.

The level of evidence is high, but the intervention has a weak recommendation in favor of adoption. More information on dosing, potential risks[337] and long-term follow-up is needed before IVIG may be widely recommended. The following issues are of particular relevance to LMICs: feasibility (e.g., the need for cold storage and for intravenous administration); cost-benefit analyses; and ensuring formulations include standardization of proper levels of antibodies against prevalent infectious agents.

The quality of evidence for using surfactant replacement therapy to prevent or treat respiratory failure, bronchopulmonary dysplasia and/or death among preterm newborns is high and this modality is used routinely in HICs. Although only three of the 50 studies reviewed are from LMICs, the intervention is strongly recommended. However, multiple challenges remain and should be addressed prior to widespread implementation: identification of optimal dosing and delivery systems, development of low-cost preparations, and refined techniques to identify high-risk newborns who may benefit from surfactant therapy.