The effect of folic acid, protein energy and multiple micronutrient supplements in pregnancy on stillbirths

We systematically reviewed all published literature to identify studies addressing peri-conceptional folic acid supplementation, balanced protein energy and multiple micronutrient supplements during pregnancy. The search strategies used for the above mentioned nutritional interventions on PubMed are given as appendices 1, 2 and 3 respectively in Additional File 1. The last date of the search was 3^rd March 2010. Initial search identified titles and abstracts relevant to the interventions of interest. Full texts were then retrieved for selected studies for final inclusion and for data abstraction. We also reviewed the reference lists of identified articles, existing reviews and meta-analyses to identify studies that were not picked up in the main search. Authors were contacted for any additional data, if required.

All randomized and quasi-randomized controlled trials assessing impact of peri-conceptional folic acid, balanced protein energy and multiple micronutrients supplements during pregnancy outcomes, were eligible for inclusion. Studies were included irrespective of language or publication status. We also reviewed observational studies for peri-conceptional folic acid supplementation as very few trials were conducted after huge protective effects of supplementation as shown in MRC trial [18]. Balanced protein energy supplementation was defined as nutritional supplementation during pregnancy in which proteins provided less than 25% of the total energy content [9]. Those studies were excluded in which the main intervention was simply dietary advice to pregnant women to increase protein intake, high protein supplementation (i.e. supplementation in which protein provided at least 25% or more of total energy content), isocaloric protein supplementation (where protein replaces an equal quantity of non-protein energy content), or low energy diets for pregnant women. Multiple micronutrients were defined as supplementation with at least five micronutrients and were compared with iron folic acid supplementation alone [7].

Every study that met the eligibility criteria was reviewed in detail and its characteristics abstracted into a standardized form [16]. The main variables extracted were country of the study, quality of allocation concealment, blinding status, characteristics of participants, sample size, description of intervention that included parameters like dose, frequency and duration of the supplements, and the follow-up period. All the studies were then graded according to the CHERG adaptation of the GRADE technique [16, 17]. Each study was allocated a quality score of ‘high’ ‘moderate’ ‘low’ or ‘very low’. This assessment was based on the methodological quality of the study and consistency of results compared to that of other selected studies [16]. Any study getting a final score of ‘very low’ was excluded from the review [16, 17]. The detailed data extraction with the limitation of studies is shown in Additional File 2.

We generated meta-analyses where data were available from more than one study. Dichotomous data were combined to get a pooled relative risk. In case where data from all the studies were not available in dichotomous form and risk ratios were available, meta-analysis was performed by generic inverse variance (GIV) method. This method is advantageous in these scenarios as it requires standard error (SE) and natural ‘log’ of effect size and data can be pooled without numerators and denominators [19]. For cluster randomized trials, we used the stated cluster adjusted relative risk and 95% confidence interval, irrespective of the method used. In case cluster adjustment was not done in the study, it was done by either adjusting the original sample size or inflating the SE times square of design effect [19]. Heterogeneity of the meta-analysis was assessed by visual inspection i.e. the overlap of the confidence intervals among the studies, and by the Chi square (P-value) of heterogeneity in the meta-analyses and I² value. A low P value (less than 0.10) or a large chi-squared statistic relative to its degree of freedom (I² >50 %) was considered as providing evidence of significant heterogeneity. In situations of substantial or high heterogeneity being present, causes were explored by sensitivity analysis. Fixed models were used for primary analysis and random models were used in case of significant heterogeneity in the pooled estimate. Results are presented as Mantel-Haenszel risk ratios (RR) and corresponding 95% confidence intervals (CI). All meta-analyses were conducted using software Review Manager Version 5 [20].

For recommendations to the LiST model, we summarized the evidence for each outcome including qualitative assessment of ‘overall’ evidence according to GRADE criteria and quantitative measures according to standard guidelines of the Child Health Epidemiological Review Group (CHERG) [16]. The qualitative evaluation of the overall (pooled) evidence was based on the volume and consistency of the evidence across studies, the size of pooled relative risk and the strength of the statistical evidence for an association between the intervention and the health outcome as reflected in the p-value [16].

The literature search identified 754 titles (Figure 1). After initial screenings of titles or abstracts and after duplicates were removed, we reviewed 26 studies. Eight studies were further excluded after review of full texts. Finally, data for eighteen studies were abstracted. One Cochrane [10] and one LiST review [21] were also available on the topic.

Out of eighteen selected studies, seven studies evaluated the effect of peri-conceptional folic acid supplementation [18, 22‐27] and eleven studies assessed fortification of food with folic acid [28‐38]. All the fortification studies were before and after studies. Of the seven supplementation studies, there were four randomized controlled trials [18, 23, 24, 26] and three cohort studies [22, 25, 27]. Four of the supplementation studies addressed primary prevention of neural tube defects [22, 23, 25, 27] and three that of secondary preventions [18, 24, 26]. Meta-analysis of four studies (1 RCT and 3 cohort studies) for primary prevention of NTDs showed a reduction of 62 % in the incidence of NTDs [relative risk (RR) 0.38; 95 % confidence interval (CI) 0.29-0.51, I squared (I²) =27.9 %, fixed model] (data not shown). The overall quality grade for this consistent pooled estimate with 1 RCT [23] and 3 cohort studies [22, 25, 27] was that of ‘moderate’ level. Disaggregated data for stillbirth due to NTDs were not available from any of these studies. Pooled results of three RCTs for prevention of recurrent NTDs showed a reduction of 70 % in recurrence of NTDs [RR 0.30; 95 % CI 0.14-0.65, I² =0 %, fixed model] (data not shown). As the evidence came from RCTs and the pooled estimate was consistent, the overall quality grade for this estimate was that of ‘high’ level. In these studies, disaggregated data for stillbirths due to NTDs were available from one study [24] with one stillbirth in the control group and no stillbirth in the intervention group, giving a relative risk of 0.17 (95 % CI 0.01-0.46).

There were eleven before and after studies that addressed primary prevention of NTDs using folic acid fortification. Pooled results from these studies showed a reduction of 41 % in primary occurrence of NTDs [RR 0.59; 95 % CI 0.52-0.68, I² =88 %, random model] (Figure 2). As there was a significant heterogeneity in the pooled estimate, random models were used. Ten of the included studies showed a clear benefit. The reason for significant heterogeneity was the difference in magnitude of ‘effect size’ of the included studies which ranged from a reduction of 26 % [33] to 78 % [38]. Although this evidence came from ‘low’ quality before and after studies, the overall quality grade for this estimate was that of ‘moderate’ level due to prominent protective effect in the same direction in most of the included studies. The disaggregated data for stillbirths due to NTDs was available from one study [31] with 5 stillbirths in intervention group and 29 in the control group giving a relative risk of 0.41 (95 % CI 0.16-1.07). This study from Canada found that before fortification (1992-1997), there were 29 stillbirths related to neural tube defects, out of a total of 531268 births, a cause-specific stillbirth rate of 5.5 per 100000 births. The corresponding figure after fortification (1998-2000) was 5 stillbirths out of 221 253, a rate of 2.3 per 100000.

Table 1 shows the summary of findings and qualitative assessment of evidence on folic acid supplementation/fortification in prevention of NTDs and related stillbirths. Keeping in mind the inadequacy of direct data on NTDs related stillbirth, we recommend an approximate reduction of 41% in NTD related stillbirths with an overall quality grade of “moderate”.

We identified 4160 titles from searches conducted in all databases (Figure 3). After screening the titles and abstracts, 22 studies were identified that addressed protein/energy supplementation during pregnancy. Eight of these studies were excluded because the intervention in these studies was either dietary advice about increase in protein/energy content, supplementation with isocaloric or high protein food [39‐46]. Fourteen studies addressed balanced protein/energy supplementation during pregnancy [47‐60]. Two of these studies were excluded because both the groups received food supplementation (high versus low energy) [53, 54]. Another study was excluded because of ‘very low’ quality [60]. Eight studies were excluded because data for outcome of interest was not available [47‐49, 51, 52, 56, 58, 59]. Finally three studies were included in the review [50, 55, 57]. A Cochrane review was also available on the topic [9]. Two of the included studies were from developing countries [50, 55] and one from a developed country [57]. The participants in all three studies were malnourished (as defined by authors). Pooled results from these three studies included a total of 2186 pregnancies and 49 stillbirths, showed that balanced protein energy supplementation during pregnancy leads to a significant reduction of 45% in all-cause stillbirths [RR 0.55, 95 % CI 0.31-0.97] (Figure 4). The overall grade quality of pooled data was that of ‘Low’ level (Table 2).

A total of 4478 titles were identified from our search strategy (Figure 5). After screening the titles and abstracts, 13 studies were selected for inclusion in this paper [61‐73]. Six of these 13 studies were cluster randomized trials [67‐69, 71, 73, 74]. Supplementation with multiple micronutrients failed to show a significant reduction in stillbirths when compared to iron folate supplementation (RR = 0.98; 95% CI: 0.88 – 1.10) (Figure 6). The impact on perinatal mortality was similar [RR 1.07, 95 % CI 0.92 –1.25] (Data not shown). As there was no significant effect of multiple micronutrient supplementation in reducing stillbirth or perinatal mortality, no recommendations have been made for LiST for this intervention (Table 3).