Impact of Double-Fortified Salt with Iron and Iodine on Hemoglobin, Anemia, and Iron Deficiency Anemia: A Systematic Review and Meta-Analysis

Abstract

Double-fortified salt (DFS) containing iron and iodine has been proposed as a feasible and cost-effective alternative for iron fortification in low- and middle-income countries (LMICs). We conducted a systematic review and meta-analysis from randomized and quasi-randomized controlled trials to 1) assess the effect of DFS on biomarkers of iron status and the risk of anemia and iron deficiency anemia (IDA) and 2) evaluate differential effects of DFS by study type (efficacy or effectiveness), population subgroups, iron formulation (ferrous sulfate, ferrous fumarate, and ferric pyrophosphate), iron concentration, duration of intervention, and study quality. A systematic search with the use of MEDLINE, EMBASE, Cochrane, Web of Science, and other sources identified 221 articles. Twelve efficacy and 2 effectiveness studies met prespecified inclusion criteria. All studies were conducted in LMICs: 10 in India, 2 in Morocco, and 1 each in Côte d'Ivoire and Ghana. In efficacy studies, DFS increased hemoglobin concentrations [standardized mean difference (SMD): 0.28; 95% CI: 0.11, 0.44; P < 0.001] and reduced the risk of anemia (RR: 0.59; 95% CI: 0.46, 0.77; P < 0.001) and IDA (RR 0.37; 95% CI: 0.25, 0.54; P < 0.001). In effectiveness studies, the effect size for hemoglobin was smaller but significant (SMD: 0.03; 95% CI: 0.01, 0.05; P < 0.01). Stratified analyses of efficacy studies by population subgroups indicated positive effects of DFS among women and school-age children. For the latter, DFS increased hemoglobin concentrations (SMD: 0.32; 95% CI: 0.03, 0.60; P < 0.05) and reduced the risk of anemia (SMD: 0.48; 95% CI: 0.34, 0.67; P < 0.001) and IDA (SMD: 0.37; 95% CI: 0.25, 0.54; P < 0.001). Hemoglobin concentrations, anemia prevalence and deworming at baseline, sample size, and study duration were not associated with effect sizes. The results indicate that DFS is efficacious in increasing hemoglobin concentrations and reducing the risk of anemia and IDA in LMIC populations. More effectiveness studies are needed.

Introduction

Anemia remains an important public health problem worldwide, primarily among women, infants, and children from low- and middle-income countries (LMICs), where traditional diets provide little bioavailable iron and where infections, especially malaria and hookworm, exacerbate the problem by increasing iron losses (1–3). Iron deficiency (ID), defined as the decrease in the total content of iron in the body, has been identified as a major cause of anemia for decades (4), and it has been linked to poor cognitive and neurologic function among children and maternal mortality and low productivity in adults, with long-term health and economic consequences (5, 6). Iron deficiency anemia (IDA) occurs when ID is sufficiently severe to reduce erythropoiesis. Because ID is an important cause of anemia, efforts to reduce its burden have been directed mostly toward increasing iron intakes through food fortification, supplementation, and dietary diversification (7).

Salt is an ideal vehicle for providing micronutrients; almost everyone consumes it and it is relatively inexpensive to fortify. The iodization of salt worldwide has been an outstanding public health success story, leading to remarkable reductions in iodine deficiency disorders (8). Several reviews and meta-analyses have examined the effects of iodine fortification on various outcomes (i.e., mental development, goiter, cretinism, iodine deficiency) (9–11) and therefore this will not be reviewed here.

One suggested way to simultaneously combat both iron and iodine deficiency disorders is through the fortification of salt with both iron and iodine, referred as double-fortified salt (DFS). DFS was first conceived of in 1969, but it took decades to develop a stable form due to technical difficulties of combining the 2 micronutrients (12). Iron microencapsulation and chelation technologies were developed to minimize iron-iodine interactions. Different types of DFS have been produced, which vary by iron and iodine compound, encapsulation of the iron or iodine, addition of additives (colorizing agents, binders, or stabilizers), and sophistication of the production method (12).

The stability, acceptability, and bioavailability of different types of DFS have been analyzed (13–24). Overall, all types of DFS have been found to be acceptable to consumers, and laboratory studies conducted have indicated good stability and bioavailability of both iron and iodine.

Several studies in various populations and contexts have examined the effects of DFS on iron and anemia; yet, to our knowledge, no formal meta-analysis of the impact of DFS on iron status has been published. We are aware of only one systematic review, which focused on cost-benefit analyses of DFS compared with other iron interventions and did not review effects on iron status (25).

The primary objective of this meta-analysis was to assess the impact of DFS on biomarkers of iron status, and the risk of anemia and IDA. The secondary objective was to assess differential effects of DFS on selected outcomes by study type (efficacy or effectiveness), population subgroups, iron formulation (ferrous sulfate, ferrous fumarate, and ferric pyrophosphate), iron concentration, duration of intervention, and study quality.

Methods

Literature search strategy

An electronic literature search was performed in MEDLINE (https://www.ncbi.nlm.nih.gov/pubmed/), EMBASE (https://www.embase.com/login), Cochrane (http://www.cochranelibrary.com), and Web of Science databases (http://apps.webofknowledge.com) to identify relevant studies that investigated the effects of DFS on iron status. The following search strategy was used for each database: (“double fortified salt” OR “dual fortified salt” OR “dual salt” OR “double salt” OR “iodine and iron”) AND (“anemia” OR “iron” OR “ferritin” OR “hemoglobin” OR “iron deficiency anemia”) AND (“trial” OR “intervention” OR “RCT” OR “Program”). The search was restricted to human studies, and no language or date restrictions were applied. The above search strategy was supplemented by a review of citations included in relevant studies and reviews. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines were followed (26).

Inclusion and exclusion criteria

The following inclusion criteria were used: 1) the study was a randomized or quasi-randomized controlled trial; 2) hemoglobin, ferritin, transferrin receptor, anemia and/or IDA were measured and reported postintervention. Studies were excluded from the meta-analysis if published reports presented insufficient information for estimating effects sizes and variances and if these could not be obtained through personal communication with the authors. Other exclusion criteria included studies analyzing the effect of salt fortified with multiple micronutrients or with iron alone and secondary analyses of original studies.

Study selection and data extraction

Two independent reviewers scanned titles and abstracts of retrieved articles and excluded irrelevant studies. Full texts of the remaining articles were reviewed to identify studies that met the inclusion criteria. If data in the original publication lacked sufficient details, the corresponding author of the study was contacted by e-mail for additional information. Reference lists of articles included and relevant literature reviews were checked for additional articles.

Retrieved studies that reported biomarkers of iron status (i.e., ferritin and transferrin receptor) presented values in a variety of ways: medians, means, and geometric means for measures of central tendency, and IQRs, SDs, and ±1 SD for measures of dispersion. Given the complexity of combining these estimates, a decision was made to focus only on hemoglobin, anemia, and IDA.

Key descriptive data, including sample size, study design, study type, duration of intervention, salt consumption, iron formulation and concentration, deworming status before the intervention, mean hemoglobin, anemia, IDA at baseline and end line in treatment and control groups, quality assessment, DFS stability, and sensory testing, were extracted into a standardized form. The reported sample size refers to the number of participants who completed the intervention. All data were entered twice and inconsistencies resolved.

Quality assessment

The Effective Public Health Practice Project (EPHPP) quality-assessment tool was used to assign a global rating to each study (27). The EPHPP assesses 8 dimensions: selection bias, study design, confounders, blinding, data collection methods, withdrawals and dropouts, intervention integrity, and robustness of the analysis. Each dimension is rated on a 3-point scale as strong, moderate, or weak, all of which contribute to the calculation of the global rating. Studies received a global rating of strong when all of the dimensions were classified as strong. Moderate and weak global scores were assigned to studies with 1 or >1 weak components, respectively. The effect of pooling results from studies of different quality was examined in subgroup analyses when possible.

Statistical analysis

The outcomes of interest were hemoglobin concentration and prevalence of anemia and IDA. Because hemoglobin was measured in a variety of ways across trials (i.e., venous blood, capillary blood, arterial blood), we used the standardized mean difference (SMD) to express the effect size (28). However, to provide the reader with a notion of mean change of hemoglobin in interpretable units (grams per liter), mean differences were also calculated for the overall pooled estimate. For anemia and IDA, we present summary effects as RRs with 95% CIs.

Postintervention hemoglobin values (means ± SDs) were used instead of changes from baseline to end line because most studies did not report the SD of the change (28).

The SMD was calculated by dividing the difference in mean hemoglobin between intervention and control groups at end line by the pooled SD. Mean differences were estimated by calculating the difference in mean hemoglobin between the intervention and control groups at end line. RRs for anemia and IDA were calculated by dividing the risk of an event in the intervention group by the risk of an event in the control group at end line. The risk of an event is estimated as the number of events divided by the total sample size.

In studies with multiple intervention groups and a single control group, the sample size of the control group was divided equally by the number of intervention groups to avoid double-counting of participants (29).

Pooled estimates for the overall effect and subgroup analyses were conducted when data were available from >1 study by using both fixed-effects and random-effects model assumptions. The random-effects model was preferred if there was evidence of significant heterogeneity. We assessed heterogeneity of effect sizes with the use of the chi-square test of homogeneity; P < 0.05 for Cochran’s Q test was considered as evidence of heterogeneity. I-square values were also examined and values >50% were deemed as exhibiting substantial statistical heterogeneity (28). We visually evaluated the presence of publication bias with the use of funnel plots (30). Sensitivity and subgroup analyses were conducted to explore predictors of positive response.

Because of the differences in coverage and compliance between efficacy and effectiveness studies, we stratified the results by study type. The specified variables for subgroup analyses included the following: 1) adult, child, and sex subgroups; 2) iron formulation; 3) iron concentration; 4) study duration; and 5) study quality. All of the statistical tests were 2-sided, with P < 0.05 reported as significant. Review Manager version 5.3 (Cochrane Collaboration) was used for the analysis of pooled estimates and preparation of forest plots and funnel plots. Effects sizes were compared informally between subgroups without performing a statistical test (28).

Random-effects meta-regression analyses were performed with the “metareg” command (STATA version 12.0; StataCorp) to explore the influence of potential effect modifiers on hemoglobin concentrations and risk of anemia. Explanatory variables included in the models were baseline hemoglobin and anemia prevalence, deworming status at baseline, sample size, and study duration.

Results

Study selection

The literature search from electronic databases and records identified through other sources yielded 221 references (Figure 1), of which 41 were duplicates and were excluded. An additional 166 studies were excluded after the title and abstract review. A total of 14 articles were assessed for eligibility, and all were included in the systematic review and meta-analysis.

FIGURE 1.

Flowchart showing the selection of trials for inclusion in the systematic review and meta-analysis.

Study characteristics and risk of bias

Twelve efficacy and 2 effectiveness studies were included. All were conducted in LMICs: 10 in India (22, 31–41), 2 in Morocco (42, 43), and 1 each in Côte D'Ivoire and Ghana (44, 45) (Supplemental Table 1). Efficacy trials provided data on 3234 intervention and 3335 control participants and effectiveness studies on 20,024 intervention and 19,402 control participants. With the use of the EPHPP quality-assessment tool test (27), 3 studies were classified as being of strong quality, 3 were of moderate quality, and 8 were of weak quality.

Pooled effects of DFS

Efficacy studies

Twelve trials reported data on the effect of DFS on hemoglobin concentrations compared with a control group, which in all studies received iodine-fortified salt. The pooled SMD for hemoglobin concentrations by random-effects model was 0.28 SD (95% CI: 0.11, 0.44; P < 0.001) (Table 1). The SMD ranged from −0.32 SD in a trial in India (22) to 1.36 SD in a study in Morocco (43) (Figure 2A). The pooled mean difference for postintervention hemoglobin concentrations by random-effects model was 3.74 g/L (95% CI: 1.42, 6.06 g/L; P = 0.002). Eight studies reported data on the prevalence of anemia and 4 studies reported on the prevalence of IDA. Pooled analysis showed significant effects of DFS on reducing the RRs of anemia (RR: 0.59; 95% CI: 0.46, 0.77; P < 0.001) and IDA (RR: 0.37; 95% CI: 0.25, 0.54; P < 0.001) (Table 1). The funnel plots were symmetrical for hemoglobin (Figure 3) and IDA (Supplemental Figure 1), which suggests an absence of publication bias in the trials included. Funnel plots for the SMD of hemoglobin of pooled efficacy and effectiveness studies were symmetrical (Supplemental Figure 2). The funnel plots were asymmetrical for anemia (Supplemental Figure 3).

TABLE 1.

Pooled and stratified estimates of efficacy studies assessing the effect of DFS in hemoglobin concentrations and risk of anemia and IDA¹

		Sample size, n				Test for heterogeneity3
Stratification variable	Trials, n	I	C	Combined effect2 (95% CI)	P	I 2, %	Q	P
Pooled effect
Hemoglobin (SMD)	12	3199	3134	0.28 (0.11, 0.44)	<0.001	89	191.09	<0.001
Hemoglobin (MD), g/L	12	3199	3134	3.74 (1.42, 6.06)	0.002	91	226.86	<0.001
Anemia (RR)	8	1469	1208	0.59 (0.46, 0.77)	<0.001	91	129.22	<0.001
IDA (RR)	4	398	433	0.37 (0.25, 0.54)	<0.001	18	4.86	0.30
Population group
Hemoglobin (SMD)
Children aged <5 y	2	382	422	−0.04 (−0.18, 0.10)	0.58	41	1.69	0.19
School-age children	7	1710	1580	0.32 (0.03, 0.60)	0.03	93	137.45	<0.001
WCA	4	414	436	0.15 (0.01, 0.28)	0.03	0	2.33	0.51
Men	2	173	166	0.10 (−0.12, 0.31)	0.37	12	1.14	0.29
Pregnant women	2	92	68	0.69 (0.36, 1.01)	<0.001	44	1.78	0.18
Anemia (RR)
School-age children	5	1214	929	0.48 (0.34, 0.67)	<0.001	95	131.21	<0.001
WCA	2	165	166	0.86 (0.70, 1.06)	0.16	31	1.44	0.23
IDA (RR)
School-age children	4	398	433	0.37 (0.25, 0.54)	<0.001	18	4.86	0.30
Iron formulation
Hemoglobin (SMD)
Ferrous sulfate	6	2222	2233	0.24 (0.01, 0.46)	0.04	92	147.47	<0.001
Ferrous fumarate	3	327	281	0.22 (0.05, 0.38)	0.01	0	1.26	0.74
Ferric pyrophosphate	3	265	212	0.64 (−0.05, 1.34)	0.07	92	25.97	<0.001
Anemia (RR)
Ferrous sulfate	3	892	722	0.69 (0.50, 0.96)	0.03	94	61.60	<0.001
Ferrous fumarate	3	328	301	0.60 (0.34, 1.06)	0.08	82	17.05	<0.001
Ferric pyrophosphate	3	249	185	0.40 (0.17, 0.94)	0.03	83	11.67	0.003
IDA (RR)
Ferric pyrophosphate	3	202	222	0.47 (0.29, 0.75)	0.001	0	1.97	0.37
Iron concentration, mg/g salt
Hemoglobin (SMD)
≤1.1	9	2794	2856	0.21 (0.04, 0.38)	0.02	89	151.45	<0.001
>1.1	3	405	278	0.56 (0.07, 1.06)	0.03	89	28.32	<0.001
Anemia (RR)
≤1.1	5	1080	947	0.74 (0.58, 0.94)	0.02	90	67.49	<0.001
>1.1	3	389	261	0.32 (0.14, 0.74)	0.007	85	20.09	0.0002
IDA (RR)
>1.1	3	334	367	0.40 (0.26, 0.61)	<0.001	21	3.78	0.29
Study duration, mo
Hemoglobin (SMD)
≤6	2	127	117	0.57 (0.12, 1.03)	0.01	68	3.12	0.08
>6	10	3072	3017	0.25 (0.08, 0.42)	0.005	90	182.42	<0.001
Anemia (RR)
≤6	2	127	117	1.31 (0.31, 5.58)	0.72	73	3.75	0.05
>6	6	1342	1091	0.54 (0.41, 0.72)	<0.001	93	129.28	<0.001
IDA (RR)
>6	3	352	384	0.30 (0.19, 0.48)	<0.001	0	1.79	0.62
Study quality4
Hemoglobin (SMD)
Strong	3	362	375	0.81 (0.18, 1.43)	0.01	94	31.48	<0.001
Moderate	2	330	195	0.30 (0.12, 0.48)	<0.001	0	0.29	0.87
Weak	7	2507	2564	0.16 (−0.01, 0.33)	0.06	86	110.59	<0.001
Anemia (RR)
Strong	2	149	154	0.50 (0.17, 1.53)	0.22	83	5.94	0.01
Moderate	2	344	215	0.34 (0.12, 0.94)	0.04	89	17.84	<0.001
Weak	4	976	839	0.72 (0.55, 0.96)	0.02	91	64.63	<0.001
IDA (RR)
Strong	2	87	94	0.23 (0.10, 0.53)	<0.001	0	0.22	0.64
Moderate	2	311	339	0.43 (0.28, 0.66)	0.0001	22	2.57	0.28

¹C, control; DFS, double-fortified salt; I, intervention; IDA, iron deficiency anemia; MD, mean difference; SMD, standardized mean difference; WCA, nonpregnant, nonlactating women of childbearing age.

²Values are SMDs, MDs, or RRs of pooled estimates, as indicated. Pooled estimates for the overall effect and subgroup analyses were conducted when data were available from >1 study.

³Random-effects models were conducted if there was evidence of significant heterogeneity; otherwise, fixed models were conducted. The inverse variance method was used for SMDs and MDs and Mantel-Haenszel was used for RRs.

⁴Studies received a global rating of “strong” when all dimensions of the Effective Public Health Practice Project quality-assessment tool (selection bias, study design, confounders, blinding, data collection methods, withdrawals and dropouts, intervention integrity, and robustness of the analysis) were classified as strong. Moderate and weak global scores were assigned to studies with 1 or >1 weak components, respectively.

FIGURE 2.

Pooled estimates of efficacy studies assessing the effect of DFS vs. control (iodized-salt) for hemoglobin concentrations (A), anemia (B), and iron deficiency anemia (C). DFS, double-fortified salt; IV, inverse variance; M-H, Mantel-Haenszel; Std., standardized.

FIGURE 3.

Funnel plot of efficacy studies for the SMD of hemoglobin (after intervention) of double-fortified salt compared with iodized salt. Values are effect sizes (SMDs) and SEs of the effect size. SMD, standardized mean difference.

Effectiveness studies

Data on hemoglobin and anemia were available from 2 effectiveness trials. The pooled SMD for hemoglobin by fixed-effects model was 0.03 SD (95% CI: 0.01, 0.05; P = 0.007). Pooled effect sizes for anemia were not significant (Table 2). The 2 effectiveness studies used ferrous sulfate for salt fortification with concentrations of 1 mg Fe/g salt, lasted >6 mo, and differed in quality (Supplemental Table 1).

TABLE 2.

Summary of pooled estimates of effectiveness studies assessing the effect of DFS on hemoglobin concentrations and prevalence of anemia¹

		Sample size, n				Test for heterogeneity3
Pooled effect	Trials, n	I	C	Combined effect2 (95% CI)	P	I 2, %	Q	P
Hemoglobin
SMD	2	20,024	19,402	0.03 (0.01, 0.05)	0.007	37	4.79	0.19
MD, g/L	2	20,024	19,402	0.40 (0.09, 0.70)	0.01	46	5.57	0.13
Anemia (RR)	2	20,024	19,402	0.97 (0.93, 1.01)	0.16	67	9.03	0.03

¹All of the effectiveness studies used ferrous sulfate with concentrations of 1 mg Fe/g salt and lasted >6 mo. C, control; DFS, double-fortified salt; I, intervention; MD, mean difference; SMD, standardized mean difference.

²Values are SMDs, MDs, or RRs of pooled estimates as indicated. Pooled estimates for the overall effect and subgroup analyses were conducted when data were available from >1 study.

³Random-effects models were conducted if there was evidence of significant heterogeneity; otherwise, fixed models were conducted. The inverse variance method was used for SMDs and MDs and Mantel-Haenszel was used for RRs.

Efficacy and effectiveness studies

The pooled effect size for hemoglobin concentrations (SMD) when effectiveness studies were included in the estimation was 0.21 SD (95% CI: 0.12, 0.29; P < 0.001). The pooled mean difference was 3.01 g/L (95% CI: 1.79, 4.24 g/L; P < 0.001) (Table 3). With regard to anemia, pooled effects of efficacy and effectiveness studies showed a significant 16% reduction in the RR of anemia for participants receiving DFS (Supplemental Figure 4). Effect sizes were not computed for IDA because effectiveness studies did not report the prevalence of IDA.

TABLE 3.

Summary of pooled and stratified estimates of efficacy and effectiveness studies assessing the effect of DFS on hemoglobin concentrations and prevalence of anemia¹

		Sample size, n				Test for heterogeneity3
Stratification variable	Trials, n	I	C	Combined effect2 (95% CI)	P	I 2, %	Q	P
Pooled effect
Hemoglobin (SMD)	14	23,223	22,536	0.21 (0.12, 0.29)	<0.001	91	296.27	<0.001
Hemoglobin (MD), g/L	14	23,223	22,536	3.01 (1.79, 4.24)	<0.001	91	294.00	<0.001
Anemia (RR)	10	21,493	20,610	0.84 (0.78, 0.92)	<0.001	87	111.3	<0.001
Population group
Hemoglobin (SMD)
Children aged <5 y	2	382	422	−0.04 (−0.18, 0.10)	0.58	41	1.69	0.19
School-age children	8	2436	2260	0.28 (0.04, 0.53)	0.02	94	161.58	<0.001
WCA	5	6691	6564	0.01 (−0.02, 0.04)	0.59	39	6.56	0.16
Men	3	5421	5315	0.04 (0.00, 0.08)	0.04	0	1.43	0.49
Pregnant women	2	92	68	0.69 (0.36, 1.01)	<0.001	44	1.78	0.18
Anemia (RR)
School-age children	6	1940	1609	0.51 (0.37, 0.70)	<0.001	94	130.41	<0.001
WCA	3	6442	6294	1.00 (0.98, 1.03)	0.83	54	4.37	0.11
Iron formulation by outcome
Hemoglobin (SMD)
Ferrous sulfate	8	22,246	21,635	0.16 (0.07, 0.25)	<0.001	92	204.18	<0.001
Ferrous fumarate	3	327	281	0.22 (0.05, 0.38)	0.01	0	1.26	0.74
Ferric pyrophosphate	3	265	212	0.64 (−0.05, 1.34)	0.07	92	25.97	<0.001
Anemia (RR)
Ferrous sulfate	5	20,916	20,124	0.92 (0.86, 1.00)	0.04	87	61.62	<0.001
Ferrous fumarate	3	328	301	0.60 (0.34, 1.06)	0.08	82	17.05	<0.001
Ferric pyrophosphate	3	249	185	0.40 (0.17, 0.94)	0.03	83	11.67	0.003
Iron concentration, mg/g salt
Hemoglobin (SMD)
≤1.1	11	22,818	22,258	0.15 (0.07, 0.24)	0.0002	90	208.03	<0.001
>1.1	3	405	258	0.56 (0.07, 1.06)	0.03	89	28.32	<0.001
Anemia (RR)
≤1.1	7	21,104	20,349	0.91 (0.85, 0.98)	0.01	84	67.88	<0.001
>1.1	3	389	261	0.32 (0.14, 0.74)	0.007	85	20.09	0.0002
Study duration, mo
Hemoglobin (SMD)
≤6	2	127	117	0.57 (0.12, 1.03)	0.01	68	3.12	0.08
>6	12	23,096	22,419	0.18 (0.10, 0.27)	<0.001	91	251.11	<0.001
Anemia (RR)
≤6	2	127	117	1.31 (0.31, 5.58)	0.72	73	3.75	0.05
>6	8	21,366	20,493	0.85 (0.78, 0.92)	<0.001	88	106.7	<0.001
Study quality4
Hemoglobin (SMD)
Strong	3	362	375	0.81 (0.18, 1.43)	0.01	94	31.48	<0.001
Moderate	3	1056	875	0.19 (−0.03, 0.41)	0.09	68	9.26	0.03
Weak	8	21,805	21,286	0.12 (0.04, 0.20)	0.002	87	143.51	<0.001
Anemia (RR)
Strong	2	149	154	0.50 (0.17, 1.53)	0.22	83	5.94	0.01
Moderate	3	1070	895	0.44 (0.23, 0.84)	0.01	84	18.97	<0.001
Weak	5	20,274	19,561	0.92 (0.86, 1.00)	0.04	86	63.61	<0.001

¹C, control; DFS, double-fortified salt; I, intervention; MD, mean difference; SMD, standardized mean difference; WCA, nonpregnant, nonlactating women of childbearing age.

²Values are SMDs, MDs, or RRs of pooled estimates, as indicated. Pooled estimates for the overall effect and subgroup analyses were conducted when data were available from >1 study.

³Random-effects models were conducted if there was evidence of significant heterogeneity; otherwise, fixed models were conducted. The inverse variance method was used for SMDs and MDs and Mantel-Haenszel was used for RRs.

⁴Studies received a global rating of “strong” when all dimensions of the Effective Public Health Practice Project quality-assessment tool (selection bias, study design, confounders, blinding, data collection methods, withdrawals and dropouts, intervention integrity, and robustness of the analysis) were classified as strong. Moderate and weak global scores were assigned to studies with 1 or >1 weak components, respectively.

Population subgroups

The SMD of hemoglobin concentration by random-effects model for school-age children was 0.32 SD (95% CI: 0.03, 0.60; P = 0.03) (n = 7 efficacy studies) (Table 1). The effect size (SMD) was 0.26 SD (95% CI: 0.08, 0.44; P = 0.005) when the 2 effectiveness studies were included into the pooled estimate (Table 3). Among women, significant effects of DFS on hemoglobin concentrations were observed for pregnant women (SMD: 0.69; 95% CI: 0.36, 1.01; P < 0.001), although only 2 efficacy studies provided data (Table 1). No significant effect of DFS on hemoglobin concentration was observed among preschool-age children (SMD −0.04; 95% CI: −0.18, 0.10; P = 0.58) (n = 2 efficacy studies).

Concerning anemia and IDA, pooled results of efficacy studies showed significant effects of DFS among school-age children, with 52% and 63% reductions in the RRs of anemia and IDA, respectively (Table 1). When effectiveness studies were included in the pooled estimate, there still was a significant reduction in the RR of anemia (49%) among school-age children who received DFS (Table 3).

Iron sources used for DFS formulation

The 3 main iron sources used for salt fortification were ferrous sulfate, ferrous fumarate, and ferric pyrophosphate. Significant and similar effects were observed on hemoglobin concentration in studies that used ferrous sulfate (SMD: 0.24; 95% CI: 0.01, 0.46; P = 0.04) (n = 6 efficacy studies) or ferrous fumarate (SMD: 0.22; 95% CI: 0.05, 0.38; P < 0.05) (n = 3 efficacy studies) (Table 1). The effect size on hemoglobin for ferric pyrophosphate was larger relative to ferrous sulfate and ferrous fumarate but had a wider CI (SMD: 0.64; 95% CI: −0.05, 1.34: P = 0.07) (n = 3 efficacy studies).

Reductions in the RR of anemia were greater for studies that used ferric pyrophosphate (RR: 0.40; 95% CI: 0.17, 0.94; P = 0.03) (n = 3 efficacy studies) relative to ferrous sulfate (RR: 0.69; 95% CI: 0.50, 0.96; P = 0.03) (n = 3 efficacy studies) and ferrous fumarate (RR: 0.60; 95% CI: 0.34, 1.06; P = 0.08) (n = 3 efficacy studies) (Table 1). Effect sizes on IDA were available for efficacy studies that used ferric pyrophosphate, showing important and significant reductions in the risk of IDA (RR: 0.47; 95% CI: 0.29, 0.75; P = 0.01) (n = 3 studies).

The 2 effectiveness studies used ferrous sulfate for salt fortification. Pooled analysis of efficacy and effectiveness studies resulted in smaller but significant effect sizes for hemoglobin concentrations (SMD: 0.16; 95% CI: 0.07, 0.25; P < 0.01) and anemia (RR: 0.92; 95% CI: 0.86, 1.00; P = 0.04) relative to pooled estimates of efficacy studies alone (Table 3).

Iron concentration of DFS

Most studies used concentrations of 1–1.1 mg elemental Fe/g salt. Only 3 efficacy studies reported DFS formulations with greater concentrations of iron. Zimmermann et al. (43) and Andersson et al. (34) reported the use of 2 mg Fe/g salt and Wegmüller et al. (45) reported the use of 3 mg Fe/g salt (Supplemental Table 1).

A significant improvement in hemoglobin concentration was observed in the stratified analysis that used DFS concentrations of >1.1 mg Fe/g salt (SMD: 0.56; 95% CI: 0.07, 1.06; P = 0.03) (n = 3 efficacy studies). Studies that used DFS iron concentrations of <1.1 mg/g salt showed a smaller but still significant effect of DFS on hemoglobin concentration (SMD: 0.21; 95% CI: 0.04, 0.38; P = 0.02) (n = 9 efficacy studies) (Table 1).

The 2 effectiveness studies used iron concentrations of ≤1.1 mg/g salt. Pooled estimates combining efficacy and effectiveness studies that used ≤1.1 mg Fe/g salt showed significant effects on hemoglobin concentration (SMD: 0.15; 95% CI: 0.07, 0.24; P = 0.0002) (n = 11 studies) and significant reductions in the risk of anemia (RR: 0.91; 95% CI:0.85, 0.98; P = 0.01) (n = 7 studies) (Table 3).

All studies that reported data on IDA were efficacy studies that used concentrations of iron >1.1 mg/g salt and showed a 60% reduction in the RR of IDA (RR: 0.40; 95% CI: 0.26, 0.61; P < 0.001) (Table 1).

Study duration

Most efficacy and all effectiveness studies lasted >6 mo (Supplemental Table 1). The effect size for hemoglobin concentrations (SMD) of efficacy studies that lasted >6 mo was 0.25 SD (95% CI: 0.08, 0.42; P = 0.005) and the RR for anemia was 0.54 (95% CI: 0.41, 0.72; P < 0.001) (Table 1).

The pooled estimate combining efficacy and effectiveness studies of >6 mo duration for hemoglobin concentration (SMD) was 0.18 (95% CI: 0.10, 0.27; P < 0.001) and for anemia the RR was 0.85 (95% CI: 0.78, 0.92: P < 0.001) (Table 3).

For studies that lasted ≤6 mo, effect sizes for hemoglobin were larger relative to studies of longer duration but had a wider CI (SMD: 0.57; 95% CI: 0.12, 1.03; P = 0.01) (n = 2 efficacy studies) (Table 1).

Study quality

For efficacy studies of strong and moderate quality, the SMDs for hemoglobin were 0.81 (95% CI: 0.18, 1.43; P = 0.01) and 0.30 (95% CI: 0.12, 0.48; P < 0.001), respectively. Efficacy studies of weak quality showed a significant effect size for anemia (RR: 0.72; 95% CI: 0.55, 0.96; P = 0.02) but had a wider CI for hemoglobin (SMD: 0.16; 95% CI: −0.01, 0.33; P = 0.06) (Table 1).

Effect modification from population baseline and study characteristics

With the use of multiple variable meta-regression analyses, we found that baseline anemia and hemoglobin values, deworming status before the intervention, sample size, and study duration were not significant predictors of the risk of anemia and hemoglobin concentration response to DFS (Supplemental Tables 2 and 3).

Stability of DFS

Seven efficacy studies presented some data on the stability of iodine, iron, and color change in DFS (Supplemental Table 1). Overall, no significant color changes were observed in DFS and control salts during transport and storage conditions, with the exception of 2 studies conducted in Morocco and Ghana (42, 44). Zimmermann et al. (42) reported that, during the dry season, when moisture content was <1%, no significant color differences between the salts were observed. However, during the damp season, when moisture content reaches ∼3%, DFS developed a mild yellow color. In the study conducted by Asibey-Berko et al. (44) in Ghana, the DFS turned a slightly darker color than the control salt.

Discussion

To our knowledge, this is the first systematic review and meta-analysis of the effects of DFS on hemoglobin status and risk of anemia and IDA. In this review of 12 efficacy and 2 effectiveness studies, we found that, among efficacy trials, DFS significantly increased hemoglobin concentrations by a mean of 3.74 g/L and reduced the RRs of anemia and IDA by 41% and 63%, respectively. Pooled estimates including the 2 effectiveness studies in addition to the efficacy trials resulted in lower but still significant effect sizes on hemoglobin but no effects on anemia.

Pooled estimates of DFS trials showed significant effects for hemoglobin concentrations, anemia, and IDA in school-age children. No effect of DFS was observed in preschool-age children in the 2 studies available. Preschool-age children have a higher potential to benefit from DFS given their increased iron demands. However, intakes of salt among preschool-age children in developing countries could be quite low because their diets usually differ from that of the rest of the family. Their food may be cooked separately and include less DFS, and consumption is highly dependent on caregiver feeding practices, which are often poor in the settings examined here (46). More studies on the effects of DFS are needed in preschool-age children.

Among women of childbearing age, pooled estimates of efficacy studies for hemoglobin concentrations were significant. Effect sizes in pregnant women were larger relative to nonpregnant women of childbearing age, but only 2 studies were available. These findings might be explained by the higher iron demands that occur in this physiologic state in which more iron is needed to support the growing fetus and placenta and to increase the maternal RBC mass (47, 48). Furthermore, there is a higher intestinal absorption of dietary iron during the second and third trimesters of gestation to accommodate for the expansion of RBC mass (49).

In terms of DFS formulations, pooled estimates of efficacy studies showed that ferrous sulfate and ferrous fumarate resulted in significant effects, with similar SMDs for hemoglobin concentrations. Estimates that included efficacy and effectiveness studies yielded similar effect sizes for ferrous fumarate and ferrous sulfate on hemoglobin concentration and the RR of anemia. Although the evidence from studies that used ferric pyrophosphate as the source of fortification (n = 3 efficacy studies) did not show a significant effect on hemoglobin concentrations, the SMD was larger than for ferrous sulfate and ferrous fumarate, and borderline significant (P = 0.07). Furthermore, the RR of IDA was reduced by 53% in studies that used ferric pyrophosphate (P = 0.001).

Some limitations should be mentioned. First, we were unable to pool results on biomarkers of iron status due to the lack of comparable statistics. Second, the unknown prevalence of infection or inflammation in the studied populations limits our ability to rely on serum ferritin and other iron biomarkers to clearly detect changes in iron status (50). Third, effect sizes were calculated on the basis of end-line values. However, all studies were randomized trials and baseline characteristics were comparable across intervention groups in most studies. Furthermore, the meta-regression showed no evidence of baseline hemoglobin concentrations and anemia prevalence predicting the risk of anemia and hemoglobin concentration response to DFS. However, the sample size was low (n = 14 studies) and the statistical power also was consequently low.

Efficacy and effectiveness of DFS

Pooling the effects of efficacy and effectiveness trials may not be the best approach for analyzing the body of epidemiologic evidence, given differences in coverage and compliance between these 2 types of studies. Efficacy trials determine whether an intervention produces the expected results under ideal circumstances, whereas effectiveness trials measure the effect under “real world” conditions. For this reason, in this meta-analysis, we present pooled estimates differentiating between efficacy and effectiveness studies in addition to combining them.

To our knowledge, only 2 effectiveness studies of DFS have been conducted (38, 39). Krämer et al. (39) undertook an evaluation of the Indian school feeding program in rural Bihar and found that DFS reduced the prevalence of anemia by 20%. Banerjee et al. (38) conducted a feasibility study of DFS purchase among households in rural Bihar. Their study examined the effectiveness of DFS sold through the local market, or distributed for free, compared with control communities and found improvements in hemoglobin concentration and a reduction in anemia prevalence among adolescents. No impact was observed for other subgroups on any of the indicators measured (anemia, cognition, mental health, and physical fitness). However, coverage was low among participants who had the option of purchasing DFS; only 14.5% of respondents were using DFS at the time of the end-line survey.

Implications for policies and programs

The direction and magnitude of the effect size for hemoglobin concentration observed in the current meta-analysis are similar to that of Gera et al. (51), who reviewed 60 trials of food fortification with iron. The study reported an overall significant improvement in hemoglobin concentrations of 4.2 g/dL (95% CI: 2.8, 5.6 g/dL). A separate review of micronutrient fortification of foods by Das et al. (52) found that iron fortification of foods resulted in a combined SMD of 0.55 (95% CI: 0.34, 0.76) for hemoglobin concentrations and an RR of 0.55 (95% CI: 0.42, 0.72) for anemia in children. In women, effects were similar for hemoglobin concentrations (0.62 SD; 95% CI: 0.36, 0.89) and for anemia (RR: 0.68; 95% CI: 0.49, 0.93). These effect sizes were larger than those reported in this meta-analysis of DFS and could be explained by a higher consumption of iron when using a combination of food vehicles such as processed foods, cereals, sauces, salt, condiments, and milk, as opposed to salt alone. However, although iron fortification of cereals is considered a low-cost intervention, it is difficult to implement on a large scale in contexts in which flour milling occurs at the village level (25). Despite these challenges, an experimental community-level iron-fortification program in Udaipur district, Rajasthan, has reported a successful local milling initiative (53). The sustainability and generalizability of such an initiative remain to be seen.

Stability and sensory testing of different DFS formulations

The stability of DFS depends on the chemical form of iron and iodine compounds, salt quality, storage conditions, and packaging material. Stability-specific studies have generally tested several DFS formulations to identify those combinations that are most stable. Currently, most DFS formulations retain >80% of their iodine content after 1 y and are well accepted by consumers (12). Stability could be further affected by the in-country salt quality and humidity (12). Changes in DFS stability have significant policy implications because of repercussions on uptake and acceptability among beneficiaries, particularly for countries with humid climates. In addition, in settings in which food is cooked for extended periods, degradation and oxidation of the iron from specific formulations could make the food unappealing.

Conclusions

Overall, these findings have important and timely implications for policies and programs that are looking to scale up DFS, as is the case for Sri Lanka, which is initiating the use of DFS in a school feeding program, as well as several Indian states that are interested in distributing DFS through the Indian food security system (i.e., ration shops). The results of this meta-analysis point to a significant potential for impact from DFS on hemoglobin concentration and a reduced risk of anemia and IDA, particularly among school-age children and women of childbearing age, including pregnant women, and begin to create the foundation to move toward more effectiveness studies.

Notes

MJR-L and LML received PhD funding from the Laney Graduate School, Emory University.

Author disclosures: MJR-L, LML, VM, and RM, no conflicts of interest.

Supplemental Tables 1–3 and Supplemental Figures 1–4 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/advances/.

Abbreviations used:

DFS

double-fortified salt

ID

iron deficiency

IDA

iron deficiency anemia

LMIC

low- and middle-income country

SMD

standardized mean difference