Effects of vitamin and mineral supplementation during pregnancy on maternal,birth,child health and development outcomes in low- and middle-income countries: A systematic review

Micronutrient and vitamin supplementation during pregnancy improves some maternal and child health outcomes

Micronutrients and vitamins are important for the body's normal functioning, growth and development, but many women and children get too few from their diets.

What is this review about?

Micronutrient deficiencies, the lack of essential vitamins and minerals, are especially a concern during pregnancy, when energy and nutrient demands are greater for both mother and foetus, and can lead to adverse health and development outcomes for the mother and child, such as being born too soon or too small.

A cost-effective strategy that has been implemented worldwide is to provide micronutrient and vitamin supplementation during pregnancy to achieve better health outcomes in the mother and child. There are different types of supplementation, including provision of a single micronutrient/vitamin (e.g. vitamin D), two micronutrients/vitamins (e.g. iron-folic acid supplementation), or several micronutrients/vitamins, which can be in the form of tablets, powders, or fat-based products.

Multiple micronutrient supplementation will help women and children who have more than one type of deficiency. However, it is understood that some micronutrients and vitamins can compete for absorption in the body and, as such, intake of several micronutrients at the same time may not be as effective as one would hope.

It is important to evaluate the various types of supplementation in pregnancy and their effects on maternal and child health outcomes to determine the best strategy for reducing poor outcomes.

What is the aim of this review?

This Campbell systematic review evaluates the various types of supplementation in pregnancy and their effects on maternal and child health outcomes to determine the best strategy for reducing poor outcomes.

What studies are included?

We searched for papers published between 1995 and 31 October 2019 in a variety of relevant databases and within grey literature. This systematic review included 314 papers across 72 studies (involving 451,723 women), of which 64 studies (involving 439,649 women) contributed to meta-analysis.

The included studies used the following comparisons: iron-folic acid supplementation versus folic-acid only; multiple-micronutrient supplementation versus iron-folic acid or iron alone; lipid nutrient-based supplementation versus multiple micronutrients; and iron, zinc, vitamin A, vitamin D or calcium supplementations, each compared to placebo.

What are the main findings of the review?

Across all comparisons, micronutrient and vitamin supplementation had little to no effect on the number of deaths among mothers and infants. Overall, women who received multiple micronutrient supplementation had fewer babies that were born early (<37 weeks of gestation), fewer babies born too small (<2500 g) and fewer babies who were smaller in size than normal for their gestational age, compared to women who received iron-folic acid supplementation. Iron or iron-folic acid supplementation was very good at improving maternal anaemia.

Mothers who received calcium supplementation may have had a decreased risk of pre-eclampsia and eclampsia during pregnancy, and mothers who received vitamin D compared to mothers who received placebo may have had fewer babies that were born early (<37 weeks of gestation).

Women supplemented with vitamin A compared to mothers given placebo had higher amounts of serum/plasma retinol, while zinc supplementation had no effect on all maternal and child outcomes reported.

Lipid nutrient-based supplementation showed little to no effect compared to multiple micronutrients; however, there were very few studies included and further research should be conducted.

What do the findings of this review mean?

The findings of this review support the use of micronutrient and vitamin supplementation during pregnancy to improve certain maternal and child health outcomes, which is especially important for populations living in low- and middle-income settings. Supplementation with multiple micronutrients was particularly good at improving several outcomes.

Further research should be done to establish the effects of supplementation on pregnant adolescents, who may have specific needs. Also, more data is needed to understand supplementation effects on long-term health and development outcomes for mother and child.

How up-to-date is this review?

The review authors searched for studies published up to October 2019.

Description of the condition

Close to two billion people today are deficient in key vitamins and minerals; of these individuals, the vast majority are women and children residing in low- and middle-income countries (LMIC) (Black et al., 2013; Development Initiatives, 2018). Among women of reproductive age (WRA) in LMIC, micronutrient deficiencies result from diets that chronically lack diversity and thus do not provide sufficient amounts of essential vitamins and minerals to meet recommended daily allowances (FAO and WHO, 2004). In some cases, infections and/or chronic disease may contribute to micronutrient deficiencies by directly inhibiting nutrient absorption (Bailey et al., 2015).

Because of increased nutritional requirements throughout pregnancy, micronutrient deficiencies are often exacerbated during this time. Additionally, repeated pregnancies and short inter-pregnancy intervals have been shown to contribute to poor maternal micronutrient status (Darnton-Hill, & Mkparu, 2015). As such, multiple concurrent deficiencies (in two or more micronutrients) are common among pregnant women, especially in LMIC (Allen & Peerson, 2009; Black, 2003). Though population-level estimates are mostly lacking, the global prevalence of prenatal iron deficiency anaemia is estimated to be 19.2% (95% confidence interval (CI) 17.1%-21.5%), while vitamin A deficiency affects 15.3% (95% CI 6.0%-24.6%) of pregnant women (Black et al., 2013). Restricting to LMIC could produce slightly larger prevalence estimates.

Micronutrient deficiencies are associated with a host of adverse outcomes for both the mother and the baby. Anemia in pregnancy, typically caused by iron deficiency, increases the risk of maternal mortality, perinatal mortality, and low birthweight (Allen 2001; Christian, 2010; Haider et al., 2013). Low folate levels are unequivocally associated with neural tube defects (NTD) (De-Regil et al., 2015], and severe iodine deficiency affects fetal development, including increasing the risk of mental retardation and cretinism (Dunn, 1993). Low calcium intake during pregnancy is associated with the development of hypertension, and hypertension is one of the leading causes of maternal morbidity, mortality, fetal growth restriction and preterm birth (Bucher et al., 1996; Ortega et al., 1999; Hofmeyr, 2018). Similarly, low vitamin D levels throughout gestation can lead to pre-eclampsia and, subsequently, increase the risk of preterm birth, small-for-gestational age (SGA), and perinatal mortality (De-Regil et al., 2016; Dror, 2011; MacKay et al., 2001). The effects of maternal zinc deficiency are not well understood, but it has been suggested that zinc supplementation during pregnancy can result in the reduction of preterm birth (Ota et al., 2015). Maternal malnutrition has also been shown to manifest through intergenerational effects, impacting the short-term and long-term outcomes of offspring, including growth, neurodevelopment and cognition, and cardiometabolic, pulmonary, and immune function (Gernand et al., 2016). Poor maternal nutrition reduces a newborn's chance to achieve proper growth and development in the short-term and, together, these early life inputs can establish the trajectory for chronic and other diseases later in life. Evidence has indicated that poor fetal and infant growth can lead to stunting in adulthood, chronic diseases relating to nutrition, lower educational attainment, reduced income, and even decreased birthweight in the next generation (Victora et al., 2008), highlighting the immense health and social consequences of maternal malnutrition.

Description of the intervention

Several strategies exist for reducing micronutrient malnutrition among women. These include diet diversification, large-scale and targeted fortification, bio-fortification of staple crops, and micronutrient supplementation with tablets or powders (Bhutta et al., 2008). This review will encompass micronutrient supplementation interventions during pregnancy.

Generally, micronutrient supplementation is used as a short term, preventive strategy that is targeted towards specific at-risk population groups (Bailey et al., 2015). As such, supplementation has been recommended as part of routine antenatal care to overcome the complications associated with micronutrient deficiencies during pregnancy.

Within the context of routine antenatal care for pregnant women, the World Health Organization (WHO) currently recomends daily iron-folic acid (IFA) supplementation with 30-60 mg of elemental iron and 400 ug of folic acid (WHO, 2016). In populations where anaemia prevalence is less than 20% or where side effects from daily supplementation are severe, weekly IFA supplementation with 120 mg of elemental iron and 2800 ug folic acid is recommended instead (WHO, 2016). The WHO has issued several context-specific recommendations as well: i) daily calcium supplementation (1.5-2.9 grams oral elemental calcium) in populations with low dietary intake of calcium; ii) daily (up to 10,000 IU) or weekly (up to 25,000 IU) vitamin A supplementation where vitamin A deficiency is a severe public health problem (WHO, 2016). Currently, zinc supplementation is recommended only where there is rigorous research to support its provision, and vitamin D supplementation is not recommended for pregnant women to improve maternal and perinatal outcomes (WHO, 2016).

To address the issue of multiple deficiencies, the United Nations Children's Fund (UNICEF), United Nations University (UNU), and the WHO developed a multiple micronutrient (MMN) tablet that provides the daily recommended intake of vitamin A, vitamin B1, vitamin B2, niacin, vitamin B6, vitamin B12, folic acid, vitamin C, vitamin D, vitamin E, copper, selenium, and iodine with 30 mg iron and 15 mg of zinc for pregnant women (UNICEF, WHO, and UNU, 1999). Other such tablets have been developed for supplementation studies on a case-by-case basis, typically providing at least three essential micronutrients.

More recently, the use of lipid-nutrient supplements (LNS) has been proposed to combat the adverse effects of maternal micronutrient deficiencies. Similar to MMN supplements, LNS contain a range of vitamins and minerals, but also provide energy, protein, and essential fatty acids. They are considered lipid-based because energy from LNS comes in the form of fats, such as vegetable fat, peanut/groundnut paste, milk powder and sugar (Arimond et al., 2015). Lipid-based products like Plumpy'nut were traditionally used for the treatment of severe acute malnutrition, but have since been adapted to contain a lower dose of energy such that daily supplementation with LNS products could be used as a preventive therapy for undernutrition (Arimond et al., 2015).

Supplementation with MMN is not recommended for pregnant women to improve maternal and perinatal outcomes, as more research is needed (WHO, 2016). The WHO has not yet issued any guidance for LNS (WHO, 2016).

How the intervention might work

Micronutrients, essential vitamins and minerals that are obtained from the diet, are critical for a host of metabolic activities that support tissue growth and functioning. As such, they are fundamental in enabling the healthy development of the fetus and promoting optimal pregnancy outcomes. Antenatal micronutrient supplementation interventions aim to increase circulating levels of vitamins and minerals in pregnant women in order to meet the recommended daily intakes, which are higher than normal due to increased physiological demands during pregnancy. Through tablets or other vehicles (e.g. syrup, drops, powder, or food matrices), the micronutrients are ingested and bioconverted to their active form in order to support maternal health and fetal development throughout gestation.

Through primary studies and meta-analysis of randomised controlled studies (RCTs), some antenatal micronutrient supplementation interventions have proven to be efficacious in improving congenital/birth outcomes, including lowering the risk of NTD, cretinism, premature rupture of membranes (PROM), low birthweight, and preterm birth (Bougma et al., 2013; De-Regil et al., 2015; De-Regil et al., 2016; Haider et al., 2013; Lassi et al., 2013; Ota et al., 2015; Rumbold et al., 2015; Zhou et al., 2013). The duration of exposure needed to produce clinically meaningful results may vary depending on the supplement. For example, it is recommended that folic acid supplementation begin as early as possible, and ideally prior to conception (WHO, 2016), while daily iron supplementation that begins mid-gestation has been effective at improving some outcomes (Peña-Rosas et al., 2015).

Why it is important to do this review

There are several existing systematic reviews that examine the impact of single and multiple micronutrient supplementation interventions in pregnancy (Appendix 1), many of which incorporate data from studies conducted in low- and middle-income settings. However, significant heterogeneity in results has been reported (e.g., for antenatal iron supplementation); this has not yet been explained by subgroup analysis. In addition, inconclusive results for several micronutrient supplementation interventions (e.g., folic acid supplementation for maternal health and pregnancy outcomes, calcium supplementation (other than for preventing or treating hypertensive disorders) for pregnancy and infant outcomes, and zinc supplementation for improving pregnancy and birth outcomes) were found, warranting further investigation. Many of the systematic reviews listed (Appendix 1) are several years old, underscoring the need to update the evidence in order to capture newly completed study data. There is the hope that with more power to detect differences, some unanswered questions will be resolved. For example, previous evidence suggests that male infants are at a greater risk for morbidity and mortality relative to their female counterparts, especially in the perinatal, neonatal and early infancy stages (Stevenson et al., 2000; Waldron, 1998; Zhao et al., 2017). As well, sex-specific differences in infant mortality following MMN supplementation were noted in a study by Smith and colleagues (Smith et al., 2017). Thus, additional exploration is required to confirm the sex-specific differences in morbidity and mortality amongst infants. Additionally, concerns have been raised regarding the safety of iron supplementation in women with high haemoglobin concentrations, and the potentially negative long-term consequences that unabsorbed iron may have on child morbidity (Mwangi et al., 2017; Paganini et al., 2016).

In addition to the limitations of existing systematic reviews of RCTs, the effectiveness of antenatal micronutrient supplementation interventions in a real world setting has not been well established. We aim to understand which antenatal supplementation interventions are effective to improve key maternal and child health, nutrition, and mortality outcomes in LMIC. We will include data from large programme evaluations as well as smaller studies. Additionally, we will include adolescent women as a pre-specified subgroup, which will help to elucidate strategies that can mitigate the risks associated with adolescent pregnancy in LMIC (Bhutta et al., 2017). Lastly, we hope to answer some of the remaining questions outlined above, including potential infant sex-specific differences and safety concerns following supplementation in pregnancy. Taken together, these results will inform the evidence on which to base policy and programming relating to micronutrient supplementation in pregnancy for women in LMIC. In addition, this review will point to any gaps in the existing evidence.

Criteria for considering studies for this review

Search methods for identification of studies

The search strategy was guided by our PICO model (Table 1), but was restricted by outcome in order to retain a broader search. The search was conducted using indexing terms, including medical subject headings (MeSH), keywords, and free text words. Details of the search strategy can be found in Appendix 2. To capture the most relevant evidence, we included articles published from 1995 to the end of June 2018 (related programmes and good quality studies before 1995 were very limited). There were no language or publication restrictions. Manual searches were conducted within reference lists of review articles and included studies, and experts were contacted to obtain any additional relevant maternal that may have been missed. The search process, including month/year of search, was documented to ensure that replication is possible.

Data collection and analysis

Description of studies

Participants

The 64 studies included in the data analyses involved 439,649 women at varying gestational ages at baseline, ranging from early pregnancy (<20 weeks of gestation) to <37 weeks of gestation. Most participants were enrolled in the studies at <or equal to 20 weeks of gestation. Included participants were healthy and without anaemia (i.e. were not recruited based on anemia status), any chronic or systemic medical condition (e.g. cardiac disease, tuberculosis, human immuno-deficiency virus (HIV)), or heightened risk of pregnancy complication (e.g. history of pre-eclampsia/eclampsia). Two studies (Ashorn et al., 2015; Friis et al., 2004) included a subgroup of pregnant women who were HIV-positive; however, the data for these subgroups were not included in this review. Across all studies, baseline characteristics of participants in the intervention and control groups were comparable, except for the following studies: Christian et al (2003); Friis et al. (2004); Ramakrishnan et al. (2003); Roberfroid et al. (2008); Zagré et al. (2007). In Christian et al., (2003), the control group included more participants who represented a specific ethnic background and owned land compared to the intervention group. In Friis et al. (2004), there were more primigravidae participants in the placebo/control group than in the intervention group. In Ramakrishnan et al. (2003), the intervention group had a higher proportion of single mothers and participants with lower mean BMI than in the control group. In Roberfroid et al. 2008, the serum hemoglobin (Hb) level of participants in the intervention group was lower than in the control group, and maternal BMI was lower in the control group than in the intervention group. Finally, in Zagré et al. (2007) the placebo group included more participants who were less educated and living in poverty than in the intervention group; as well, there were more households and preventive measures against malaria amongst participants in the intervention group.

Intervention

All supplements were given orally and in the form of tablets (except for lipid-based supplements). Supplements were given to pregnant women throughout the remainder of their pregnancy from the time of enrolment.

IFA Supplementation versus Folic Acid Supplementation

Seven studies assessed IFA supplementation compared to folic acid supplementation or placebo (Christian et al., 2003; Etheredge et al., 2015; Liu et al., 2013; Menendez et al., 1995; Zeng et al., 2008; Zhao et al., 2015; Ziaei et al., 2007). Of these, 5 were included in the iron versus placebo comparison, with folic acid supplementation provided as a co-intervention (Christian et al., 2003; Etheredge et al., 2015; Liu et al., 2013; Menendez et al., 1995; Ziaei et al., 2007). Three studies had multiple intervention arms and were included in other comparisons: Christian et al., 2003; Liu et al., 2013; Zeng et al., 2008. Studies provided 30 mg to 60 mg of iron, with the majority providing 60 mg of iron, and 400-500 ug of folic acid in their supplements.

MMN supplementation versus IFA supplementation or placebo

Thirty-four studies assessed MMN versus IFA supplementation (Aminisani et al., 2009; Asemi et al., 2016; Ashorn et al., 2015; Bhutta et al., 2009; Caulfield et al., 1999; Choudhury et al., 2012; Christian et al. ,2003; Dewey, 2009; Dijkhuizen et al., 2001; Fawzi et al., 2007; Friis et al., 2004; Hafeez et al., 2005a, 2005b; Hanieh et al., 2013; Huy et al., 2009; Kæstel et al., 2005; Liu et al., 2013; Merialdi et al., 2004; Moore et al., 2012; Muslimatun et al., 2001; Osrin et al., 2005; Ramakrishnan et al., 2003; Roberfroid et al., 2008; Roth et al., 2013 (AViDD); Roth et al., 2018 (MDIG); Saaka et al., 2009; Semba et al., 2001; Sorouri et al., 2016; SUMMIT Study Group, 2008; Sunawang et al., 2009; Tofail et al., 2008; Villar et al., 2006; West et al., 2014; Zagré et al., 2007; Zeng et al., 2008).

For this review, MMN was defined as the provision of at least 3 micronutrients (e.g. iron, folic acid and vitamin A). Trials could be considered MMN even in situations where iron with or without folic acid was provided in a separate supplement. Thus, it is possible that these studies could be included in other comparisons as well. For example, where a study investigates the effects of vitamin A versus placebo supplementation, but all women are provided IFA as the standard of care, then data from this study would be included in the vitamin A versus placebo and MMN vs. IFA comparisons. There were twelve of these types of studies (Aminisani et al., 2009; Caulfield et al., 1999; Dijkhuizen et al., 2001; Hafeez et al., 2005a, 2005b; Merialdi et al., 2004; Muslimatun et al., 2001; Roth et al., 2013 (AViDD); Roth et al., 2018 (MDIG); Saaka et al., 2009; Semba et al., 2001; Sorouri et al., 2016; and Villar et al., 2006) that were included in other comparisons; mainly zinc versus placebo and vitamin A versus placebo. Asemi et al. (2016) evaluated calcium with vitamin D supplementation versus placebo (plus IFA), and was included in the MMN versus IFA comparison; however, this study was not included in any other comparison given the insufficient number of studies to undertake meta-analyses for calcium with vitamin D supplementation. Five studies had multiple intervention arms and were included in other comparisons: Ashorn et al., 2015; Christian et al., 2003; Dewey, 2009; Liu et al., 2013; Moore et al., 2012; Zeng et al., 2008.

The composition of MMN supplement varied across all studies. Eleven studies used the UNIMMAP formulation, developed by UNICEF, WHO and United Nations University (Bhutta et al., 2009; Huy et al., 2009; Kæstel et al., 2005; Liu et al., 2013; Osrin et al., 2005; Roberfroid et al., 2008; SUMMIT Study Group, 2008; Sunawang et al., 2009; Tofail et al., 2008; Zagré et al., 2007; Zeng et al., 2008). Another 3 studies used an adapted UNIMMAP formulation which contained the exact same combination of vitamins and minerals, but in different dosages (Hanieh et al., 2013; Moore et al., 2012; West et al., 2014). The remaining 20 studies used non-UNIMMAP formulations for their MMN supplements (Aminisani et al., 2009; Asemi et al., 2016; Ashorn et al., 2015; Caulfield et al., 1999; Choudhury et al., 2012; Christian et al., 2003; Dewey, 2009; Dijkhuizen et al., 2001; Fawzi et al., 2007; Friis et al., 2004; Hafeez et al., 2005a, 2005b; Merialdi et al., 2004; Muslimatun et al., 2001; Ramakrishnan et al., 2003; Roth et al., 2013 (AViDD); Roth et al., 2018 (MDIG); Saaka et al., 2009; Semba et al., 2001; Sorouri et al., 2016 and Villar et al., 2006).

The dose of iron in the MMN supplement differed across all studies. The studies that used UNIMMAP all contained 30 mg of elemental iron (Bhutta et al., 2009; Huy et al., 2009; Kæstel et al., 2005; Liu et al., 2013; Osrin et al., 2005; Roberfroid et al., 2008; SUMMIT Study Group, 2008; Sunawang et al., 2009; Tofail et al., 2008; Zagré et al., 2007; Zeng et al., 2008). Of the studies that used an adapted UNIMMAP formulation, 2 studies used 60 mg of iron (Hanieh et al., 2013; Moore et al., 2012) and 1 study used 27 mg of iron (West et al., 2014). Of the studies that used a non-UNIMMAP formulation for the MMN supplement, 7 studies included 60 mg of iron in their formulation (Caulfield et al., 1999; Choudhury et al., 2012; Christian et al., 2003; Fawzi et al., 2007; Merialdi et al., 2004; Muslimatun et al., 2001; Ramakrishnan et al., 2003). Seven other studies used dosages of iron <60 mg, ranging from 20 mg to 40 mg (Aminisani et al., 2009; Ashorn et al., 2015; Dewey et al., 2009; Dijkhuizen et al., 2001; Saaka et al., 2009; Semba et al., 2001; Sorouri et al., 2016), while 2 studies used 66 mg of iron (Roth et al., 2013 (AViDD); Roth et al., 2018 (MDIG). Two studies (Friis et al., 2004; Hafeez et al., 2005a, 2005b) did not indicate the dose of iron in the MMN supplement.

Lipid-based nutrient supplementation versus MMN supplementation or placebo (control)

Four studies assessed LNS supplementation versus MMN supplementation or placebo (Ashorn et al., 2015; Dewey et al., 2009; Huybregts et al., 2009; Moore et al., 2012). Ashorn et al. (2015) and Dewey, 2009 utilized the same formulation for the LNS supplement, presented in a powder form: 20 g sachets containing 2.6 g protein, 10 g fat, 4.59 g linoleic acid, 0.59 g linolenic acid. These were given alongside a non-UNIMMAP formulation of multiple micronutrients. Huybregts et al. (2009) also provided a powder that contained 1.56 MJ of energy: carbohydrates 15.9 g, protein 14.7 g, fat 27.6 g, MUFA 12.1 g, PUFA 7.3 g, Omega-3 fatty acids 0.4 g, omega-6 fatty acids 7.0 g), 9.1 g total dietary fibre; this was given alongside a UNIMMAP MMN supplement. Moore et al. (2012) provided LNS that contained energy (746 kcal), protein (20.8 g), lipids (52.6 g), alongside a non-UNIMMAP MMN supplement.

Vitamin A versus placebo

Nine studies assessed vitamin A supplementation versus placebo (Cox et al., 2005; Darling et al., 2017; Kirkwood et al., 2010; Muslimatun et al., 2001; Prawirohartono et al., 2011; Semba et al., 2001; Tanumihardjo, 2002; West et al., 1999; West et al., 2011). Of these, 1 study was excluded from analysis (Prawirohartono et al., 2011). Studies provided vitamin A in doses ranging from 10,000 IU of retinol weekly to 70,000 IU of retinol weekly, with the majority of studies providing 17,000-25,000 IU of retinol weekly.

Zinc versus placebo

Thirteen studies assessed zinc supplementation versus placebo (Ahmad et al., 2016; Aminisani et al., 2009; Castillo-Durán et al., 2001; Caulfield et al., 1999; Christian et al., 2003; Darling et al., 2017; Dijkhuizen et al., 2001; Hafeez et al., 2005a, 2005b; Merialdi et al., 2004; Osendarp et al., 2000; Prawirohartono et al., 2011; Saaka et al., 2009; Sorouri et al., 2016). Of these, 1 was excluded from analysis (Prawirohartono et al., 2011). Studies provided zinc in doses ranging from 15 mg to 50 mg of zinc sulphate daily. The majority of studies provided 20 to 30 mg of zinc daily.

Iron versus placebo

Thirteen studies assessed iron supplementation versus placebo (Christian et al., 2003; Etheredge et al., 2015; Falahi et al., 2010; Korkmaz et al., 2014; Liu et al., 2013; Menendez et al., 1995; Ouladsahebmadarek et al., 2011; Preziosi et al., 1997; Tanumihardjo, 2002; Zeng et al., 2008; Zhao et al., 2015; Ziaei et al., 2007, 2008). Of these, 1 was excluded from analysis (Korkmaz et al., 2014). Studies provided iron in doses ranging from 30 mg to 100 mg. The majority of studies provided 50 or 60 mg of iron; one study (Ouladsahebmadarek et al., 2011) provided 30 mg of iron and Preziosi et al., 1997 provided 100 mg.

Vitamin D versus placebo

Eleven studies assessed vitamin D supplementation versus placebo (Asemi et al., 2013; Hossain et al., 2012; Khan et al., 2016; Mohammad-Alizadeh-Charandabi et al., 2015; Naghshineh & Sheikhaliyan, 2016; Roth et al., 2013 (AViDD); Roth et al., 2018 (MDIG); Sabet et al., 2012; Sablok et al., 2015; Sahu et al., 2009; Vaziri et al., 2016). Studies provided doses of vitamin D ranging from 2800 IU to 75,000 IU of vitamin D weekly. The majority of studies provided vitamin D in doses ranging from 10,000 to 25,000 IU per week.

Calcium versus placebo

Five studies assessed calcium supplementation versus placebo (Belizán et al., 1997; Jarjou et al., 2006; Kumar et al., 2009; López-Jaramillo et al., 1997; Villar et al., 2006). Of these, 1 study was excluded from being pooled with other studies in meta-analysis (Jarjou et al., 2006). Studies provided calcium supplementation in doses ranging from 500 mg daily to 2000 mg of calcium daily.

Excluded studies

We excluded 1,364 studies from this review. Of these, 434 studies were excluded based on an ineligible intervention; 466 were of the wrong study design; 353 involved the wrong population (e.g. high-income country or child population); 75 reported ineligible outcomes; 17 involved the wrong comparison; 5 were abstracts; and 14 studies were inaccessible. Please see the Characteristics of excluded studies for the excluded studies found through our grey literature and hand-searches, and their respective reasons for exclusion (n = 295). For the remaining excluded studies (n = 1069) found through the various database searches and their respective reasons for exclusion, please refer to Tables 5–16.