Interventions to Improve Adolescent Nutrition: A Systematic Review and Meta-Analysis

Adequate adolescent nutrition is an important step for optimal growth and development. In this article, we systematically reviewed published studies till December 2014 to ascertain the effectiveness of interventions to improve adolescent nutrition. We found one existing systematic review on interventions to prevent obesity which we updated and conducted de novo reviews for micronutrient supplementation and nutrition interventions for pregnant adolescents. Our review findings suggest that micronutrient supplementation among adolescents (predominantly females) can significantly decrease anemia prevalence (relative risk [RR]: .69; 95% confidence interval [CI]: .62–.76) while interventions to improve nutritional status among “pregnant adolescents” showed statistically significant improved birth weight (standard mean difference: .25; 95% CI: .08–.41), decreased low birth weight (RR: .70; 95% CI: .57–.84), and preterm birth (RR: .73; 95% CI: .57–.95). Interventions to promote nutrition and prevent obesity had a marginal impact on reducing body mass index (standard mean difference: −.08; 95% CI: −.17 to .01). However, these findings should be interpreted with caution due to significant statistical heterogeneity.

Adequate adolescent nutrition is an important step for optimal growth and development. In this article, we systematically reviewed published studies till December 2014 to ascertain the effectiveness of interventions to improve adolescent nutrition. We found one existing systematic review on interventions to prevent obesity which we updated and conducted de novo reviews for micronutrient supplementation and nutrition interventions for pregnant adolescents. Our review findings suggest that micronutrient supplementation among adolescents (predominantly females) can significantly decrease anemia prevalence (relative risk [RR]: .69; 95% confidence interval [CI]: .62e.76) while interventions to improve nutritional status among "pregnant adolescents" showed statistically significant improved birth weight (standard mean difference: .25; 95% CI: .08e.41), decreased low birth weight (RR: .70; 95% CI: .57e. 84), and preterm birth (RR: .73; 95% CI: .57e.95). Interventions to promote nutrition and prevent obesity had a marginal impact on reducing body mass index (standard mean difference: À.08; 95% CI: À.17 to .01). However, these findings should be interpreted with caution due to significant statistical heterogeneity. Adolescent nutrition is crucial for proper growth and development and a prerequisite for achieving full developmental potential. Suboptimal nutrition may contribute to delayed and stunted growth [1] as well as impaired development. As adolescents undergo a period of rapid growth and development, adequate nutrient intake (of both macro and micronutrients) is critical. Many of the risk factors that impact maternal and newborn health exist right from adolescence, including nutritional deficiencies. Prepregnancy wasting in adolescents is usually reflected as low body mass index (BMI < 18.5). Low BMI significantly increases perinatal risks including stillbirths, preterm births, small for gestational age, and low birth weight (LBW) babies [2]. Iron deficiency anemia is among the top 10 causes of disability-adjusted life years lost among adolescents [2]. Concern is especially warranted for adolescent girls because their iron requirements are relatively high (due to growth spurts, sexual maturation, and menstrual losses) and because they may be on the cusp of motherhood. While most programs are targeted at pregnant women, the depletion of iron stores in women starts during adolescence with the onset of menstruation. More recently, there has been a growing interest in adolescent girls' nutrition as a means to improve the health of women and children. Each year around 16 million babies are born to adolescent girls between the ages of 15 and 19 years, accounting for over 10% of the total births each year [3]. Pregnancy in adolescence is associated with greater risk to the mother and newborndincluding anemia, mortality, stillbirths, and prematuritydespecially since the adolescent girls are not physically mature themselves [3]. Adolescent girls are two to five times more likely to die from pregnancy-related causes than women aged 20e29 years [3]. Girls younger than 19 years have a 50% increased risk of stillbirths and neonatal deaths, as well as an increased risk for preterm birth, LBW, and asphyxia [3]. These health risks further increase for girls who become pregnant earlier than 15 years and are somewhat reduced for older adolescents aged 18e19 years.
Over the last two decades, increasing rates of overweight and obesity among children and adolescents have been observed in many countries [4,5]. Many low-and middleincome countries (LMICs) now bear a double burden of nutritional disorders due to the emerging issue of overweight and obesity along with the existing high rates of stunting and other micronutrient deficiencies [6,7]. Childhood overweight is associated with multiple immediate and long-term risks including raised cholesterol, raised triglycerides, type 2 diabetes, high blood pressure, adult obesity, and its associated consequences [8,9]. Prepregnancy overweight has been linked to two of the foremost causes of maternal mortality (hypertensive disorders of pregnancy and gestational diabetes mellitus) [10e13] as well as other adverse pregnancy outcomes, including poor lactation practices [14,15], obstetric anesthesiaerelated complications [16], prolonged gestation [17,18], maternal infectious morbidity [19], and decreased success with trials of labor.
This article is part of a series of reviews conducted to evaluate the effectiveness of potential interventions for adolescent health and well-being. Detailed framework, methodology, and other potential interventions have been discussed in separate articles [20e26]. In this article, we systematically reviewed published literature to ascertain the effectiveness of interventions to promote nutrition among adolescents comprising of micronutrient supplementation, nutrition interventions for pregnant adolescents, and interventions to prevent obesity.

Methods
For the purpose of this review, the adolescent population was defined as aged 11e19 years; however, since many studies targeted youth (aged 15e24 years) along with adolescents, exceptions were made to include studies targeting adolescents and youth. Studies were excluded if they targeted age groups other than adolescents and youth or did not report segregated data for the age group of interest. Searches were conducted till December 2014, and we did not apply any limitations on the start search date or geographical settings. Outcomes were not prespecified, and we included all the outcomes reported by the study authors. We searched systematically for existing reviews and took a systematic approach to consolidate the existing evidence through the following methodologies: 1. De novo review: For interventions where no reviews existed, we conducted a new review; and 2. Updating existing reviews: We updated the existing systematic reviews only if the existing review included evidence before 2011.

Methodology for de novo reviews
For de novo reviews, our priority was to select existing randomized, quasi-randomized and before/after studies, in which the intervention was directed toward the adolescent age group and related to nutritional outcomes. A separate search strategy was developed for each aspect using appropriate keywords, Medical Subject Heading, and free text terms. The following principal sources of electronic reference libraries were searched to access the available data: The Cochrane Library, Medline, PubMed, Popline, LILACS, CINAHL, EMBASE, World Bank's Jolis search engine, CAB Abstracts, British Library for Development Studies at Institute of Development Studies, the World Health Organization regional databases, Google, and Google Scholar. The titles and abstracts of all studies identified were screened independently by two reviewers for relevance and matched. Any disagreements on selection of studies between these two primary abstractors were resolved by the third reviewer. After retrieval of the full texts of all the studies that met the inclusion/ exclusion criteria, data from each study were abstracted independently and in duplicate into a standardized form. Quality assessment of the included randomized controlled trials (RCTs) was done according to the Cochrane risk of bias assessment tool. We conducted meta-analysis for individual studies using the software Review Manager, version 5.3 (Cochrane Collaboration, London, United Kingdom). Pooled statistics were reported as the relative risk (RR) for categorical variables and standard mean difference (SMD) for continuous variables between the experimental and control groups with 95% confidence intervals (CIs). A grade of "high," "moderate," "low," and "very low" was used for grading the overall evidence indicating the strength of an effect on specific health outcome according to the Grading of Recommendations Assessment, Development and Evaluation criteria [27].

Methodology for updated reviews
We updated the existing systematic reviews only if the most recent review on a specific intervention was conducted before December 2011. For updating the existing reviews, we adopted the same methodology and search strategy mentioned in the existing review to update the search and find all the relevant studies after the last search date of the existing review. After retrieval of the full texts of all the articles that met the inclusion/exclusion criteria, data from each study were abstracted independently and in duplicate into a standardized form. Information was extracted on study design, geographical setting, intervention type and description, mode of delivery, and outcomes assessed. We then updated the estimates of reported outcomes by pooling the evidence from the new studies identified in the updated search and reported new effect size for the outcomes of interest with 95% CIs. We then assessed and reported the quality of included reviews using the 11-point assessment of the methodological quality of systematic reviews criteria [28].

Results
Based on our search results, we updated one systematic review and conducted two de novo reviews.
For the impact of "micronutrient supplementation among adolescents" and "nutrition interventions for pregnant adolescents," we conducted de novo reviews (as there were no relevant existing reviews) while for interventions to prevent obesity, we updated an existing Cochrane review by Waters et al. [29]. Figure 1A describes the search flow, and the characteristics of the included studies for the de novo reviews are detailed in Table 1.
The outcome quality for micronutrient supplementation was rated as "moderate quality" because for various outcomes there was considerable heterogeneity; and generalizability was limited to females because most of the studies included female participants. A summary of quality of evidence is provided in Table 2.
The quality of outcomes for interventions for pregnant women was rated to be moderate to low due to the study design limitations, heterogeneity, and limited generalizability of the interventions ( Table 3). The quality of outcomes for promoting healthy nutrition and preventing obesity was rated to be of "moderate" quality due to design limitation, heterogeneity, and limited generalizability to high-income countries (HICs) only (Table 4).

Micronutrient supplementation for adolescents
A total of 31 studies were included, of which 23 were conducted in LMICs [30e59, 76]. Studies evaluated the effectiveness of iron, folic acid, vitamin A, vitamin D, vitamin C, calcium, zinc, and multiple micronutrients supplementation to adolescent population. Thirteen studies evaluated the impact of iron/iron folic       on adolescent girls except for nine studies that included adolescent boys and girls. Findings from moderate-quality evidence suggest an overall significant reduction in anemia (as defined by study authors) with iron/iron folic acid supplementation alone or in combination with other micronutrient supplementation (RR: .69; 95% CI: .62e.76; Figure 2). Subgroup analysis according to the delivery settings suggests that school-based delivery significantly reduced anemia (RR: .67; 95% CI: .60e.74) while evidence from community-based delivery was underpowered. School-based delivery of iron/iron folic acid supplementation alone or in combination with other micronutrient supplementation was also associated with improved serum hemoglobin (mean difference [MD]: 1.94 g/dl; 95% CI: 1.48e2.41), ferritin (MD: 3.80 mcg/L; 95% CI: 2.00e5.59), and iron (MD: 6.97 mmol/L; 95% CI: .19e13.76).
Zinc supplementation led to improved serum zinc concentrations (MD: .96 mcg/dl; 95% CI: .81e1.12) while calcium and vitamin D supplementation did not have a clear impact on vitamin D levels and parathyroid hormone. Gender-specific subgroup analysis suggests significant improvements in both genders; however, most of the studies were conducted on adolescent girls.

Nutrition interventions among "pregnant adolescents"
A total of 16 studies were included outlining interventions intended to modify maternal diet and reduce adverse maternal and perinatal outcomes. The study participants were lowincome, pregnant adolescents from prenatal clinics in urban areas in Chile, Ecuador, United States of America, or Canada, between the ages of 13 and 20 years. All included studies were clinic based except for one school-based [72] and one community-based [69] study. The intervention commenced between 20 and 27 weeks of gestation and continued until delivery. The intervention strategies mainly involved provision of micronutrient supplementation such as calcium and zinc, in addition to the routine iron folic acid supplementation to adolescent mothers or engaging them in nutritional education sessions to enable them to improve nutritional intake. Long-term nutritional counseling was frequently employed whereby pregnant adolescents would have access to a nutritionist whom they would consult as part of antenatal care. Pooled data from moderate-to low-quality evidence suggested a statistically significant improvement in mean birth weight (SMD: .25; 95% CI: .08e.41; Figure 3), reduced LBW (birth weight < 2500 g; RR: .70; 95% CI: .57e.84; Figure 4), and preterm birth (before 37 weeks; RR: .73; 95% CI: .57e.95). These results must be interpreted with caution due to high heterogeneity and very small impact.

Promoting healthy nutrition and preventing obesity
We updated the existing Cochrane review on promotion of healthy nutrition and preventing obesity by Waters et al. [29] for the age group 11e19 years with an assessment of the methodological quality of systematic reviews rating of 11 for the update. A total of 10 studies (five from the existing review þ five new studies) from HICs were included. Overall, the impact on BMI was marginally significant (SMD: À.08; 95% CI: À.17 to .01; Figure 5). Further subgroup analysis revealed that physical activity or dietary control alone did not have any significant impact on BMI reduction while school-based delivery strategies were found to be more effective than interventions in noneducational settings.

Discussion
Our review suggests that micronutrient supplementation among adolescents can significantly decrease the prevalence of anemia in this age group with school-based supplementation having significant impact while evidence from communitybased studies was found to be underpowered. It must be noted, however, that most of these studies were centered on female adolescents and did not take into account the male adolescent population hence limiting the generalizability of the intervention. Most studies on adolescent micronutrient supplementation were conducted in LMICs making the findings context specific and relevant since these settings bear most of the global burden of undernutrition and micronutrient deficiencies. The impact of individual micronutrients and gender-specific impacts could not be segregated. Furthermore, included studies targeted overlapping age groups among the adolescent population that might lead to variations in the outcome effect. However, these findings should be interpreted with caution due to high heterogeneity. Interventions to improve nutritional status of pregnant adolescents significantly improved neonatal birth weight, decreased LBW, and preterm birth. However, there were insufficient data to evaluate the impact on perinatal, maternal, and neonatal mortality. However, these findings should be interpreted with caution due to high heterogeneity. These findings have limited generalizability since all the findings were from HICs. There is a need to implement the proven interventions in LMIC settings with a higher burden of undernutrition and food insecurity. A focus on adolescent girls' nutrition is important not only to improve the health status of women but also to ensure optimal fetal growth and development to prevent the vicious cycle of intergenerational transmission of undernutrition. Further studies evaluating safety and potential long-term impact of such interventions and cost-effectiveness of these strategies are needed.
Our review suggests that interventions to promote nutrition and prevent obesity can marginally reduce BMI. Evidence from the interventions to prevent obesity mostly comes from HICs hence limiting the generalizability of findings to HIC settings only. With the increasing trend of childhood obesity in LMICs, there is a need for future studies on obesity prevention in LMIC settings [77e79]. Furthermore, these countries have much higher rates of LBW babies and stunting, and a consequent higher risk of adulthood obesity. There is strong evidence to suggest that once the adolescent is obese, it may be difficult to reverse, with obesity continuing through adulthood, strengthening the case for primary prevention and specific focus on LMICs [80,81].
Existing reviews on nutrition promotion and obesity prevention have overlapping age groups and include children, adolescents, and youth. Our findings are in concordance with other reviews that suggest beneficial impacts of programs that combine the promotion of healthy dietary habits and physical activity on preventing obesity in children and adolescents, especially school-based programs [82,83]. Furthermore, evidence exists that a combination of interventions including nutrition, physical activity, knowledge, attitudes, or healthrelated behaviors has the potential to reduce the risk factors associated with obesity among preadolescent girls (7e11 years), although the sustainability of the effects of such interventions is less clear [84]. Some studies also highlighted important barriers to increasing physical activity among girls including lack of suitable places, resources, and social support for physical activity hence limiting compliance with the intervention program [29].
Limitations of our review include high heterogeneity, lack of data from LMICs, and lack of data to conduct gender-specific subgroup analysis. There is sufficient evidence suggesting the importance of adolescent nutrition interventions and its impact on improved adolescent nutrition and birth outcomes. Countries should now specifically focus on this age group and design programs accordingly with a greater focus on reaching out to this vital segment of the population through schools. There is a need to adopt multisectoral approach in targeting the adolescent age group involving schools and communities through policies and programs to improve adolescent nutrition. Future studies, especially obesity related, should focus on LMICs and underprivileged populations in HIC settings to have maximum impact on improving adolescent nutrition status and reducing adverse neonatal outcomes in these settings. There is also a need to investigate the association of improved adolescent nutrition with improved cognition and future productivity.

Funding Sources
The preparation and publication of these papers was made possible through an unrestricted grant from the Bill & Melinda Gates Foundation (BMGF).