Micronutrient Fortified Condiments and Noodles to Reduce Anemia in Children and Adults—A Literature Review and Meta-Analysis

Micronutrient deficiencies impose a considerable burden of disease on many middle and low income countries. Several strategies have been shown to be effective in improving micronutrient deficiencies. However, the impact of fortified condiments as well as fortified noodles is less well documented. We aimed to investigate existing evidence on the impact of micronutrient fortified condiments and noodles on hemoglobin, anemia, and functional outcomes in children and adults (age: 5 to 50 years). We conducted a literature review in electronic databases. In addition, we screened the homepages of relevant organizations and journals. We included randomized controlled trials (RCT). Of 1046 retrieved studies, 14 RCT provided data for the meta-analysis. Micronutrient fortification of condiments and noodles increased hemoglobin concentrations by 0.74 g/dL (95%-confidence intervals (95%-CI): 0.56 to 0.93; 12 studies) and 0.3 g/dL (95%-CI: 0.12 to 0.48; 1 study), respectively. Micronutrient fortification also led to a reduced risk of having anemia (risk ratio 0.59 (95%-CI 0.44 to 0.80)). Ferritin concentrations increased with fortified condiments. Functional outcomes were rarely assessed and showed mixed results. The use of micronutrient fortified condiments can be a strategy to reduce anemia in children and adults due to micronutrient deficiencies. The effect of fortified noodles seems to be smaller.


Introduction
Micronutrient (MN) deficiencies impose a considerable burden of disease on many middle-and low-income countries, resulting for example in reduced growth, high anemia prevalence, or increased infection rates [1]. Several strategies, proposed in recommendations and guidelines, have been shown to be effective in preventing or reducing MN deficiencies in different target populations and with different carriers for MN [2][3][4]: Examples include food-based approaches (such as spreads to increase energy-density and MN content of food; MN powders for home fortification with sprinkles; fortified condiments) and MN supplementation (such as vitamin A capsules administered at defined intervals). Fortification of staple food (such as fortified salt, noodles, flour or oil) is also widely used to resolve MN deficiencies in general populations.
Distribution of fortified food via government programs does not always work well due to logistical problems or inappropriate priority setting. Thus, commercially distributed fortified foods that fit local nutrition habits may be an additional option to improve nutritional status in developing countries. Fortified condiments (widely consumed in Africa and Asia) and fortified noodles (specifically consumed in Eastern-Asia) are two such centrally processed nutrition types which are already on the market, even in low-income countries.

Inclusion/Exclusion Criteria
To answer our research question, we defined the following inclusion criteria: Population: Children and adults from 5 to 50 years. Intervention: Micronutrient fortified condiments or noodles products. Fortified condiments include: condiments, salt, seasonings, soy sauce, fish sauce, bouillon, sprinkle, powder. Micronutrients for fortification include: iron, vitamins, zinc, iodine, folate, calcium, phosphorus, magnesium, selenium. Control intervention: Non-fortified condiments and noodles. Outcome: Serum markers with direct health impact (such as hemoglobin; anemia rate), functional status, quality of life, morbidity (as measured with physical or mental health measures), or mortality. We also included socioeconomic outcomes, productivity, or acceptability of fortified products. Design: Randomized controlled trials, RCT. We excluded studies with children under 5 years and adults over 50 years. Nutritional intervention solely based on supplementation, salt fortified only with iodine, or interventions to test the bioavailability of micronutrients were excluded. Fortified staples, oils, or herbs were also not included. Change in surrogate parameters (e.g., MN blood concentrations) was not an included outcome. Surrogate parameters can be relevant from a nutrition science perspective, but they do not by themselves facilitate a comparison of the direct health impact of the intervention.

Study Selection and Data Extraction
In advance, we conducted training sessions to increase consistency between reviewers. Two reviewers screened titles and abstracts for relevance. Unclear cases were discussed between reviewers, and if necessary, with a third reviewer. Disagreements were resolved by consensus.
Potentially relevant studies were assessed by full text according to the pre-specified inclusion and exclusion criteria. A data extraction form was developed using Microsoft Excel, pilot tested on a small selection of studies and adjusted as necessary. Data were extracted by one reviewer and checked independently by a second reviewer.
The following information was extracted: study identifying items (such as author, year, location, and setting), population details (such as number of participants, age, sex, exclusion criteria), intervention details (such as carrier of MN, type of fortification, and concentration), results (such as numerical data for effectiveness/Hb and Ferritin, adverse events). If several papers reported results from the same population, each population was included only once for analysis.
To assess risk of bias of the included studies, quality assessment forms were developed on Microsoft Excel using current guidelines for the conduct of interventional studies [12]. We included the following domains: randomization (generation of random sequence; allocation concealment); blinding of participants; incomplete outcome data; and selective outcome reporting. According to pre-specified criteria for risk-of-bias assessment, data were extracted for each included RCT. For example, outcome data were deemed complete, if follow-up data were provided for at least 80% of participants or missing outcome data had been imputed (for details of other criteria, please see study protocol). We used the following categories: YES (criterion fulfilled), NO (criterion not fulfilled) and "?" (unclear, not enough information given). Risk-of-bias assessment was carried out by one reviewer. Unclear cases were discussed with a second reviewer. Disagreements were resolved by consensus.

Analysis
We performed an analysis with statistical pooling for continuous variables like hemoglobin changes (forest plots of weighted mean differences and 95%-confidence intervals (CI)). If means and standard deviations of changes were not reported, we calculated change as the difference between baseline and final values for the intervention and control group and applied the SD of final values [7]. For anemia rates, we calculated risk ratios and 95%-CI. We calculated the heterogeneity between the analyzed studies with I 2 , which is the percentage of the variation in the estimated effects due to heterogeneity rather than by chance [14]. I 2 lies between 0% and 100%. A rough guide to interpretation is as follows: Heterogeneity of 0% to 40% might not be important; a range of 30% to 60% may represent moderate, 50% to 90% substantial, and 75% to 100% considerable heterogeneity. We tested the robustness of results by calculating weighted mean differences for hemoglobin concentrations at the end of follow-up.
Furthermore, we divided our studies into reasonable subgroups, as pre-specified in our protocol, and performed subgroup analyses depending on the data available. Domains for subgroups comprised fortification strategy (single MN strategy vs. dual/multi MN strategy), region (Asia vs. Africa), food carrier (condiments vs. noodles), type of iron salt (NaFeEDTA vs. other salts) and risk of bias (low vs. intermediate/high). Studies with low risk of bias were defined as fulfilling at least four out of five quality domains ("YES"). An ex-post subgroup analysis compared targeted populations (children or adolescents vs. other populations (mostly women in childbearing age; sometimes entire communities)).
Finally, we performed a meta-regression analysis weighted for the inverse of the variance of the outcome to assess the influence of single parameters on hemoglobin change [14]. Such parameters were hemoglobin concentrations before intervention, follow-up completeness, and length of follow-up.
Significance p-values of <0.05 were used. Analyses were carried out with STATA SE 12.1 software package (Stata-Corp. 2011. Stata Statistical Software, College Station, TX, USA).
The carrier used for micronutrients was salt six times [21,[24][25][26][27][28], fish sauce three times [15][16][17], soy sauce twice [32,33], and seasoning [30], noodles [18], and masala powder once [20]. Only in two trials was multi-micronutrient (MMN) fortification [28,30] used, e.g. additional fortification with vitamin A, zinc, and folate. In the other 12 studies [15][16][17][18]20,21,[24][25][26][27]31,33], iron was the net difference in MN exposition between intervention and control groups. For example, some studies used a dual MN strategy with iron and iodine in the intervention group and mono-iodized salt in the control group, while other studies used a single MN strategy with iron in the intervention group and non-fortified food in the control group. Finally, we performed a meta-regression analysis weighted for the inverse of the variance of the outcome to assess the influence of single parameters on hemoglobin change [14]. Such parameters were hemoglobin concentrations before intervention, follow-up completeness, and length of follow-up.
Significance p-values of <0.05 were used. Analyses were carried out with STATA SE 12.1 software package (Stata-Corp. 2011. Stata Statistical Software, College Station, TX, USA).
The carrier used for micronutrients was salt six times [21,[24][25][26][27][28], fish sauce three times [15][16][17], soy sauce twice [32,33], and seasoning [30], noodles [18], and masala powder once [20]. Only in two trials was multi-micronutrient (MMN) fortification [28,30] used, e.g. additional fortification with vitamin A, zinc, and folate. In the other 12 studies [15][16][17][18]20,21,[24][25][26][27]31,33], iron was the net difference in MN exposition between intervention and control groups. For example, some studies used a dual MN strategy with iron and iodine in the intervention group and mono-iodized salt in the control group, while other studies used a single MN strategy with iron in the intervention group and non-fortified food in the control group.    The 14 studies with 15 comparisons (one study provided two comparisons between four different groups [34]) comprised 8845 people with a mean age between 7.3 and 34 years. Follow-up periods were mostly under one year (mean follow-up: 0.88 years; range: 2.4 months to 2 years). Most participants lived in a rural region. In two trials, participants were women of childbearing age. The most common setting for recruitment was the school setting (nine trials). The populations included in the primary studies were representative of the target population for fortified food, i.e., MN deficiencies were common. However, persons with severe disease (e.g. anemia with Hb < 8g/L), were often excluded from the RCTs. Adherence to intervention was not reported.

Effects on Hemoglobin Concentrations
Hemoglobin concentrations in blood was the most frequently reported outcome parameter. Median baseline hemoglobin concentrations were 11.8 g/dL (IQR: 11.1 to 12.5) for intervention groups and 12.0 g/dL (IQR: 11.0 to 12.4) for control groups. Micronutrient fortification of condiments and noodles (mostly iron as the only difference between compared groups) increased hemoglobin concentrations by 0.68 g/dL (95%-CI: 0.51 to 0.85; I 2 = 96%; 13 RCT with 14 comparisons and 8845 participants) compared to control groups ( Figure 2).
The 14 studies with 15 comparisons (one study provided two comparisons between four different groups [34]) comprised 8845 people with a mean age between 7.3 and 34 years. Follow-up periods were mostly under one year (mean follow-up: 0.88 years; range: 2.4 months to 2 years). Most participants lived in a rural region. In two trials, participants were women of childbearing age. The most common setting for recruitment was the school setting (nine trials). The populations included in the primary studies were representative of the target population for fortified food, i.e., MN deficiencies were common. However, persons with severe disease (e.g. anemia with Hb < 8g/L), were often excluded from the RCTs. Adherence to intervention was not reported.

Effects on Hemoglobin Concentrations
Hemoglobin concentrations in blood was the most frequently reported outcome parameter. Median baseline hemoglobin concentrations were 11.8 g/dL (IQR: 11.1 to 12.5) for intervention groups and 12.0 g/dL (IQR: 11.0 to 12.4) for control groups. Micronutrient fortification of condiments and noodles (mostly iron as the only difference between compared groups) increased hemoglobin concentrations by 0.68 g/dL (95%-CI: 0.51 to 0.85; I 2 = 96%; 13 RCT with 14 comparisons and 8845 participants) compared to control groups ( Figure 2). Comparison of pre-specified subgroups in different domains showed no statistically different effect on hemoglobin concentrations for fortification strategy (single MN strategy vs. dual/multi MN strategy) and region. Hemoglobin concentrations showed a more pronounced increase in the 11 studies with high risk of bias (0.78 g/dL; 95%-CI: 0.47 to 1.08) compared to the two studies (three comparisons) with low risk of bias (0.41 g/dL; 95%-CI: 0.26 to 0.57), but again, the difference was not statistically different. Condiments showed a higher impact on hemoglobin change (increase of 0.74 g/dL; 95%-CI: 0.56 to 0.93) than noodles (increase of 0.3 g/dL; 95%-CI: 0.12 to 0.48), but data for noodles were from one single study. Finally, different types of iron preparations showed no differences in rise of hemoglobin concentrations (NaFeEDTA: 0.69 g/dL vs. 0.68 g/dL for other Comparison of pre-specified subgroups in different domains showed no statistically different effect on hemoglobin concentrations for fortification strategy (single MN strategy vs. dual/multi MN strategy) and region. Hemoglobin concentrations showed a more pronounced increase in the 11 studies with high risk of bias (0.78 g/dL; 95%-CI: 0.47 to 1.08) compared to the two studies (three comparisons) with low risk of bias (0.41 g/dL; 95%-CI: 0.26 to 0.57), but again, the difference was not statistically different. Condiments showed a higher impact on hemoglobin change (increase of 0.74 g/dL; 95%-CI: 0.56 to 0.93) than noodles (increase of 0.3 g/dL; 95%-CI: 0.12 to 0.48), but data for noodles were from one single study. Finally, different types of iron preparations showed no differences in rise of hemoglobin concentrations (NaFeEDTA: 0.69 g/dL vs. 0.68 g/dL for other preparations).
Also our ex-post subgroup analysis showed no relevant difference between children/adolescents and other targeted populations.

Effects on Anemia Prevalence
For the definition of anemia, most studies relied on the WHO definition [35] and used thresholds between 11 g/dL and 13 g/dL, depending on age and gender of the investigated population. The median of anemia rates at baseline was 46% (IQR: 26% to 95%). Six studies reported iron deficiency anemia rates based on ferritin concentrations (median: 55% (IQR: 38% to 77%) in this subgroup). Again, in most of the studies, iron fortification was the only difference between intervention and control groups.
The risk of having anemia in the intervention groups compared to control groups was 0.59 (95%-CI: 0.44 to 0.80; I 2 = 83%) in 10 RCT (11 comparisons; Figure 3). Similar anemia rates emerged from the comparison of studies with high and low risk of bias (high risk: 0.58; 95%-CI: 0.42 to 0.81, low risk: 0.63; 95%-CI: 0.36 to 1.1). Also in the five other subgroup domains (fortification strategy; region; food carrier; type of iron salt; targeted populations) no significant differences emerged for anemia rates.

Effects on Anemia Prevalence
For the definition of anemia, most studies relied on the WHO definition [35] and used thresholds between 11 g/dL and 13 g/dL, depending on age and gender of the investigated population. The median of anemia rates at baseline was 46% (IQR: 26% to 95%). Six studies reported iron deficiency anemia rates based on ferritin concentrations (median: 55% (IQR: 38% to 77%) in this subgroup). Again, in most of the studies, iron fortification was the only difference between intervention and control groups.
The risk of having anemia in the intervention groups compared to control groups was 0.59 (95%-CI: 0.44 to 0.80; I 2 = 83%) in 10 RCT (11 comparisons; Figure 3). Similar anemia rates emerged from the comparison of studies with high and low risk of bias (high risk: 0.58; 95%-CI: 0.42 to 0.81, low risk: 0.63; 95%-CI: 0.36 to 1.1). Also in the five other subgroup domains (fortification strategy; region; food carrier; type of iron salt; targeted populations) no significant differences emerged for anemia rates.

Other Reported Effects
Outcomes other than hemoglobin concentrations, anemia rates, or ferritin concentrations are rarely assessed in the primary studies. Five studies additionally reported morbidity outcomes, two studies assessed cognition and one trial described productivity outcome. Among these few studies, no consistent pattern of relevant effects of micronutrient fortification of condiments or noodles emerged on outcomes like infections and anthropometric parameters. Consumer acceptability of salt was not influenced by fortification.
One RCT from India [28] compared multi-micronutrient salt with iodized salt in schoolchildren and found a significantly higher memory score (ability to memorize words in the same sequence as played by audiotape before) in the intervention group than in the control group.
One randomized trial from Thailand [29] looked at the cognitive function (short term learning, memory, attention span, visual recall) of primary-school children with or without iron-fortified salt. After 31 weeks, they found a significant intervention effect of iron-fortified salt only on visual recall.
Another trial from India [26] reported the productivity of adult tea pickers with iron-fortified salt compared to no fortified salt. After one year, the average amount of tea picking was significantly higher in the intervention group than in the control group.

Results of Meta-Regression
The multivariable meta-regression analysis showed no significant association between the independent variables (hemoglobin concentrations before intervention, follow-up completeness, and length of follow-up) and the change in hemoglobin concentrations.

Risk of Bias Assessment
Only two trials provided enough information to classify random sequence generation and allocation concealment as adequate [18,26] (Table 3). The other 12 studies did not provide enough information to assess if procedures were performed adequately. Blinding of participants was mentioned in most of the trials. Seven studies addressed incomplete outcome data and 10 trials showed no selective outcome reporting (for example besides serum markers, also anemia, height/weight were reported).
In summary, the risk of bias for the investigated outcomes "hemoglobin change" and "anemia rates" is unclear in most of the studies and should be considered in the final conclusions.

Discussion
In our review with meta-analysis we studied hemoglobin, anemia, and functional outcomes of MN fortified condiments and noodles for children and adults. In summary, MN fortification (single MN strategy with iron or dual/multiple MN strategy with iron) showed a relevant increase in hemoglobin concentrations and decrease in anemia rates mainly for condiments. In our data set of 14 RCT, functional outcomes were rarely assessed (for example, morbidity, cognition, and productivity outcomes) and studies showed mixed results. Methodologic shortcomings of the primary studies were common and should be considered in the final conclusions. For example, details of randomization procedure (generation of random sequence; allocation concealment) were often not reported and it thus remains unclear if randomization was performed according to recommended standards. In addition, adherence to intervention was often not reported.

Existing Systematic Reviews and Research Needs
Other systematic reviews (SR) have evaluated the health effects of MN interventions. These SRs [5,9,[36][37][38][39] differ from our review, either regarding the carrier (no condiments or contemporaneous analysis with other carriers) or because of the included designs of the studies (such as non-RCTs). Four of these SRs investigated the impact of MN fortified food on hemoglobin concentrations or anemia prevalence [5,9,36,39]. They all found an increased hemoglobin concentration or reduced risk of anemia due to fortified food with either iron [5,9,39], vitamin A and iron [5], or multiple micronutrients [5,36]. The other two reviews included primary studies with data for outcomes beyond hemoglobin concentrations and anemia rates, for example goiter prevalence or birth weight [37,38].
In our study, no difference in the increase of hemoglobin concentrations between fortification strategies (single/dual/multiple MN strategy) was found. This is somewhat surprising as other studies, which included mostly children and adolescents, showed an increased effect of multiple MN fortification compared to single MN fortification [6,7]. Possible explanations for this disagreement might be differential iron carriers or different target populations, as we included age groups from 5 to 50 years.
Standards for methodological rigor of primary studies in the nutrition field are an important research need as primary studies are the evidence base for reviews. The challenges and limitations of this type of research are highlighted by the result of our risk of bias assessment which addresses internal validity. For example, future studies in this field should adhere to established reporting standards for randomization procedure to give reviewers a better understanding if a possible selection bias has led to an overestimation of effect. Concerning external validity, more information about the representativeness of study participants for the real life target population is needed. For example, little socio-economic data was provided to estimate how well the study participants matched the profile of the most needy population groups of the country. Affordability is of outstanding importance to judge any public health effect of commercially distributed fortified food.

Implications for Decision Makers
Fortification of food and in particular condiments can be delivered in different ways. For example iodine fortified salt is a best practice example of a mandatory regulation by government in selected countries [40]. Salt is a widely used and low cost carrier for fortification. Other options like commercially distributed fortified food are also already available in many markets, even in low-income countries. A limitation of this market access is that the very poor population may not be reached. Each country has its own culture, economy, and social characteristics which will affect the delivery and management of fortified food to their population [41]. Thus, it may not be helpful to develop a single distribution strategy for all countries.
To better understand the impact of fortified food in "real world" settings, the monitoring of health effects is important. The Working Group [42] used a reasonable follow-up of 18 months for their large-scale effectiveness trial conducted in the general population of three rural areas and one urban area in India. The hemoglobin concentrations and anemia rates improved significantly in the intervention communities of this field trial after iron-fortification of salt. The change in Hb after 12 months ranged between 0.25g/dL and 3.28g/dL according to different areas, sex, and age. The prevalence of anemia before the intervention ranged between 6.5% and 99.2% and after the intervention between 6.5% and 60.9% according to different areas, sex, and age. Furthermore, in a recent review about food fortification in India observational studies provide evidence of positive health effects of salt fortification programs in real world settings [43].
Additional economic analyses should be performed to better understand the health economic effects of fortified condiments and noodles and support decision makers in their policy.

Strengths and Limitations
A strength of this review is the systematic approach adhering to critical methodological issues for synthesis [12] and reporting of evidence [13]. Even though we cannot be sure we have found all relevant studies, we are convinced that we found enough relevant studies to provide reliable data in this review.
There are also several limitations to be mentioned. First, the risk of bias in the included studies is unclear, due to poor reporting of the randomization procedure in the primary studies. Second, the statistical heterogeneity between the analyzed trials was high, thus pooled estimates have to be interpreted carefully. Third, we did not screen all references independently by two reviewers and did not perform independent, double data extraction or assessment of risk of bias. This could have led to bias. Fourth, we could have missed some studies because of the broad field of possible search terms for "condiments". Finally, as only one noodle study was retrieved, there is insufficient aggregated evidence to conclude whether noodles by themselves improve health outcomes.

Conclusions
Micronutrient fortified condiments can be a strategy to reduce anemia in children and adults due to micronutrient deficiencies beyond supplementation programs and fortification of staple food. The effect of fortified noodles seems to be smaller, but conclusions are based on one study only. Risk of bias in the included studies is unclear and should be considered in the final conclusion.
Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6643/8/2/88/s1, Figure S1: Effect of fortified condiments and noodles on ferritin levels compared to non-fortified condiments or noodles. Included are 3 studies with suitable ferritin data for meta-analysis. Results are provided as weighted mean difference in ferritin (WMD: micro-g/L with 95%-CI) between intervention and control group.
Author Contributions: Sascha Hess designed and conducted research, analyzed data, and drafted the manuscript. Klaus Eichler helped to design research, conducted research, helped to analyze the data and to draft the manuscript. Linda Tecklenburg conducted research. All authors read and approved the final manuscript.