Trends in Micronutrient Interventions, Anemia, and Iron Deficiency among Women and Children in Guatemala, 2009–2019

Background Food fortification and micronutrient supplementation are public health strategies to improve micronutrient status in Guatemala; their population effectiveness has not been evaluated in recent years. Objective We evaluated trends in food fortification, micronutrient supplementation, anemia, and iron deficiency among nonpregnant women aged 15–49 y [women of reproductive age (WRA)] and children 6–59 aged mo [preschool age children (PSC)]. Method Nationally representative serial cross-sectional surveys were used to assess changes in hemoglobin, anemia, ferritin, iron deficiency, iron deficiency anemia, and self-reported consumption of fortifiable foods and micronutrient supplements during 2008/2009, 2013, 2015, 2016, 2017/2018, and 2018/2019. Predictors of hemoglobin and ferritin were assessed using generalized linear mixed models adjusted for survey year as random effects, and the consumption of fortifiable foods, supplements, and other potential confounders were fixed effects. Results Multiple micronutrient powder consumption among PSC during the previous 3 mo was 53.3% (95% CI: 49.4, 57.2) in 2013 and 33.6% (28.8, 38.4) in 2018/2019. Anemia among PSC was 11.3% (8.0, 14.5) in 2008/2009 and 6.1% (3.6, 8.6) in 2018/2019. Anemia among WRA was 10.7% (7.2, 14.2) in 2008/2009 and 3.9% (2.7, 5.2) in 2018/2019. Iron deficiency among PSC was 15.5% (12.1, 19.0) in 2008/2009 and 10.9% (7.4, 14.5) in 2016 (lowest), but 17.1 (13.3, 21.0) in 2017/2018 (highest). Iron deficiency among WRA was 14.9% (11.6, 18.2) in 2008/2009, 13.8% (11.8, 15.8) in 2013 (lowest), and 18.9% (16.3, 21.6) in 2017/2018 (highest). Wheat flour/bread consumption was positively associated with hemoglobin among PSC, and sugar consumption was positively associated with hemoglobin among WRA. The reported consumption of fortifiable foods was not associated with ferritin among PSC or WRA. Conclusions Guatemala has implemented multiple food fortification strategies, and anemia has declined. Increases in iron deficiency in 2017–2019 warrant further attention. Secular trends toward poverty alleviation, education, and development might be responsible for changes not explained by the micronutrient interventions evaluated.


Introduction
Affecting an estimated 1.74 billion people worldwide, anemia is a major cause of morbidity globally resulting in an estimated 58.6 million years lived with disability in 2019 [1]. The condition results from multiple etiologies, including both micronutrient deficiencies and nonnutritional causes, such as illness or blood loss [2]. Iron deficiency is a common cause of anemia in most populations, although causes vary by population [3]. Vitamins A, B12, and folate are other micronutrient causes, among others [4,5]. Anemia is associated with poor health and economic outcomes, and related micronutrient deficiencies have additional negative consequences, which makes addressing them a public health priority [4,6].
Guatemala currently mandates fortification of salt and 4 staple foods: salt, sugar, wheat flour, and nixtamalized maize flour (Supplemental Table 1), the last one was added to Guatemala's fortification portfolio in 2016 by the government mandate, meaning that 2 staple foods (wheat and maize flour) have been mandatorily fortified with iron [7]. A popular brand of sugar in Guatemala voluntarily began fortifying with iron (ferric amino acid chelate) in late 2008 [8]. Specialty cereal products, including some provided by the government, are also voluntarily fortified with iron [9]. Guatemala's efforts to strengthen enforcement of its fortification mandates are well documented and include the establishment of the National Commission for the Fortification, Enrichment, and/or Equalization of Foods, which coordinates and supervises food fortification programs [9].
Because, in general, young children do not consume industrially fortified foods in sufficient quantity, micronutrient supplementation is also used for prevention in this population. In Guatemala, high-dose vitamin A supplements were administered biannually to preschool-aged children 6-59 mo from the early 2000s to 2014, after which it was determined that there was sufficient evidence of adequate year-round vitamin A intake and low deficiency to limit biannual supplementation to only those aged 6-24 mo [10], and it was further limited to those aged 6-11 mo in 2018 [11]. Since 2010, multiple micronutrient powders (MNPs) containing iron, folic acid, zinc, copper, selenium, iodine, and vitamins A, B1, B2, B3, B6, B12, C, D, and E have been distributed for home fortification of foods given to children aged 6-59 mo. Unlike mass fortification, home fortification requires more intensive interventions at the consumer level aimed at behavior change and demand generation [12].
National nutrition surveys in Guatemala before 2013 showed a gradual decline in anemia, although these surveys occurred with long intervals of 5-10 y between them and only one in 2008/2009 evaluated micronutrient deficiencies. These national surveys did not evaluate industrially produced fortifiable food consumption and micronutrient content of fortifiable foods. Changes in indicators and methodologies also create additional issues with comparing these data over time. To resolve these issues, the Integrated Reproductive Health and Nutrition Surveillance System (Sistema de Vigilancia Epidemiol ogica de Salud y Nutrici on [SIVESNU]) was created [13].
Using the national micronutrient survey from 2008/2009 (Encuesta Nacional de Micronutrientes, ENMICRON) as an anchor and SIVESNU serial cross-sectional survey data from 5 cycles between 2013 and 2019, we conducted a secondary data analysis of trends in anemia, micronutrient status, micronutrient supplementation, and fortification. Our study was guided by a program impact pathway for population-based fortification programs [14]. The impact pathway outlines key indicators for monitoring and evaluating food fortification programs. Indicators include the availability, quality, coverage within households, consumption by individuals, and micronutrient contribution of fortified foods [14]. We hypothesized that fortification programs in Guatemala resulted in staple foods with higher iron, vitamin A, and other micronutrient contents, and the consumption of fortifiable foods and micronutrient supplements was associated with higher levels of hemoglobin and serum ferritin concentration during the 10-y period.

Questionnaires
An adult household member reported information on household demographics and expenditures. Household food expenditures focused on the quantity and types of sugar, bread, and maize flour. Separate questionnaires collected information from WRA and primary caregivers of PSC on demographics and consumption of fortifiable foods and micronutrient supplements. The fortifiable food consumption was assessed in terms of type and frequency of consumption by the individual in the home within the previous 24 h and 7 d. The individual consumption of fortifiable foods was expressed as the number of days the food was consumed over the previous 7 d categorized as 0, 1-2, and 3-7 d. Fortifiable foods were categorized into the following 4 groups: 1) sugar or sugar-sweetened foods or beverages, 2) bread or wheat flour, 3) fortified cereals including Incaparina and VitaCereal, and 4) maize flour (not including tortillas made outside the home). Household poverty was assessed by the Global Multidimensional Poverty Index (MPI) and categorized into 4 levels of intensity: 1) no poverty, 2) vulnerable to poverty, 3) poverty, and 4) severe poverty [15]. The Global MPI has been used in Guatemala and other low-and middle-income countries as an objective and comparable measure of poverty. Current use of reversible contraceptives among WRA, dichotomized as current or noncurrent users, included the use of intrauterine device, contraceptive injection or implant, oral contraceptive, vaginal contraceptive method, or hormonal contraceptive patch in the previous 30 d.

Food samples
Samples of bread and sugar were collected in each of the SIVESNU cycles. Complete details of the sample collection, processing, and analysis of food samples have been published previously [19][20][21][22][23]. A sample of bread was collected from the local bakery or store where bread was reported to be most often purchased by households within each cluster. Bread was not collected from households, whereas standard (white) sugar was collected from a random subsample of households. Vitamin A content was assessed by the quantitative spectrophotometric method. Iron was measured in sugar samples by acid microwave digestion, followed by atomic emission spectrometry. Iron in bread and retinol in sugar were compared with regulatory minimums: 33 and 5 mg/kg, respectively [36,37]. Iron in sugar was compared with the 6 mg/kg that producers reported as the minimum level of voluntary fortification reported by the manufacturer [8].

Statistical analyses
Descriptive statistics (means and percentages) were calculated for characteristics of WRA and PSC within each survey year. The household-level coverage of 6 fortifiable foods, the presence in the household, quantity purchased, and the country of origin were described using medians with interquartile range or percentage. Across the years, differences in the population prevalence were examined using survey design-adjusted Rao-Scott chi-square tests, and differences and trends in mean continuous variables were tested using design-adjusted linear regression. Kruskal-Wallis tests were used to examine nutrient content of fortifiable foods from bakeries and households in terms of iron in bread and sugar, and retinol in sugar, which were compared across the survey years. Crude prevalence estimates across survey years of anemia and anemia-related micronutrient deficiencies were visualized with plots.

Trend analyses and model covariates
Overall trends in anemia, Hb, iron deficiency, (log-transformed) ferritin, and iron deficiency anemia were assessed separately by survey linear regression (identity link) with survey year as the sole fixed effect [38]. We applied complex survey linear regression models to study trends in Hb and ferritin, and their associations with participant-, household-, and population-level covariates over the survey years. In these multivariate generalized linear mixed effects models (GLMMs), adjusted secular trends were captured as survey year with a random intercept. Participant-level characteristics were modeled as fixed effects of the studied population during each year. Model covariates (confounders and effect modifiers) were selected based on prior literature and included age, parity (WRA only), sex (PSC only), ethnicity, rurality, remittances (WRA only), education (WRA only), BMI, LAZ/HAZ (PSC only), household poverty, RBP, current use of reversible contraceptives (WRA only), frequency of fortifiable food consumption, and micronutrient supplementation (PSC only). Because of the nearly universal consumption of sugar in the surveys, only standard sugar (azúcar est andar), fortified with vitamin A and potentially fortified with iron, was used as a covariate in GLMMs.
Associations between the covariates and nutritional indicators are presented in g/dL for Hb and percentage change for ferritin. Average survey year (random slopes) effects were calculated as best linear unbiased predictors (BLUPs). The use of BLUP estimates enabled the adjustment for the intersurvey and intrasurvey variations in biomarker distributions and are interpreted as the difference in population mean Hb or ferritin from the grand marginal mean (intercept) of all participants over all survey years [39]. Bootstrapped estimates of 500 iterations were used to generate 95% CIs around the BLUPs.
Given potential differences in fortification levels across the years (effect modification), separate GLMMs containing a statistical interaction term of fortifiable food consumption and survey year were used to assess differential effects.
Sampling weights were applied where indicated. Statistical significance was set at P value of < 0.05 with 95% CI as needed. All analyses were conducted in SAS 9.4 (SAS Institute).

Basic characteristics
The 6 nationally representative surveys have a combined sample size of 4133 PSC and 9008 WRA. These samples represent 900,000 PSC and 4 million WRA during each year of the study period [40]. Table 1 compares selected demographic and nutrition-related characteristics of PSC across the study period. The distribution of age, sex, ethnicity, and rurality were consistent across time (P ¼ 0.08, 0.98, 0.69, and 0.08, respectively). Poverty decreased over the studied period (P < 0.001). For example, the proportion experiencing poverty and severe poverty were 27.  Table 2 compares selected demographic and nutrition-related characteristics among WRA. The distribution of age and the proportion of indigenous women were unchanged across the surveys (P ¼ 0.80 and 0.85, respectively). Parity decreased, whereas the number of years of schooling increased (P < 0.001). Poverty decreased (P < 0.001), and the distribution of BMI shifted toward obesity (P ¼ 0.001). For example, 18.6% had obesity in 2008/2009 and 25.9% had obesity in 2018/2019. Figure 1 describes the level of the fortification of bread with iron and of sugar with iron and vitamin A with regulatory and reported minimums shown as reference lines. The median concentration of iron in bread varied significantly across the study period (P < 0.001) from a low in 2013 of 50.7 mg/kg to a high in 2015 of 55.4 mg/kg. All bread samples, except 2, exceeded regulatory minimums, suggesting that the wheat flour used to make bread was adequately fortified with iron (data not shown). The median concentration of iron in sugar also varied across the study period (P < 0.001) from a high in 2013 of 3.3 mg/kg to a low in 2016 of 0.9 mg/kg. The intrinsic value of iron in standard (white) sugar in Guatemala is between 0.1 and 1 mg/kg [41], and the manufacturer reported the minimum level of iron in fortified sugar is 6 mg/kg [8].

Fortifiable foods in households
Supplemental Table 2 describes fortifiable staples found within Guatemalan households. Sugar consumption, especially standard unrefined sugar, is ubiquitous within households. However, there has been a significant decrease in the presence of standard sugar and corresponding increase in the presence of brown sugar within households from 2013 to 2018/2019 (both P < 0.001). Among labeled sugar, nearly all were national brands. Sweetened and sandwich bread were present in~13% and 10% of households, respectively, across all surveys. Nearly all (>95%) bread was unlabeled. Approximately 2 units of bread were purchased per household per week. Maize flour was present in~8% of households, potentially representing how households purchase prepared maize tortillas rather than raw flour. More than 90% of maize flour was labeled, and most were national brands.

Frequency of consumption of fortifiable foods
Supplemental Figure 1A and B shows the reported weekly consumption of sugars, breads, fortified cereals, and maize flour consumed by PSC and WRA, respectively, in their homes from 2013 to 2018/2019. Sugar consumption was nearly universal among both populations: >95% within each year consumed it 3-7 d/wk. Bread/wheat flour is also commonly consumed, and the distribution of frequency of consumption per week has increased (P ¼ 0.008) with 84% of PSC consuming it at least once per week in 2013 and 2015 and 91% consuming it as often in 2017/2018 and 2018/2019. Among WRA, the distribution of reported bread/wheat flour consumption also shifted toward more frequent consumption (P < 0.001), with 88% consuming it at least once per week in 2013 and 95% consuming it as often in 2018/2019. The consumption of specialty fortified cereals varied over time among PSC (P ¼ 0.01) and among WRA (P < 0.001). Fifty-two percent of PSC and 38% of WRA reported consuming specialty fortified cereals at least once per week in 2013, and these proportions increased to 58% and 49%, respectively, in 2018/ 2019. Reported consumption of maize flour in homemade foods increased significantly only for PSC (P ¼ 0.03), although weekly consumption was low for both populations (10%-15%).  Supplemental Figure 2A and B shows the prevalence of anemia-related micronutrient deficiencies where available at each survey among PSC and WRA, respectively. Among PSC, the prevalence of elevated sTfR, a marker of iron deficient erythropoiesis, was 4.0% (2.4, 5.5)  At the only measurement, there was no folate deficiency among PSC or WRA in 2008/2009 (data not shown). Based on limited data from 2008/2009, 2013, and 2016, zinc deficiency was above or just below the cut-point for classification as a problem of public health concern (>20%) in both populations [42]. The prevalence of low vitamin B12 among PSC was 18.9% (13.8, 24. Table 3 shows potential factors associated with Hb and ferritin among PSC in Guatemala. A higher frequency of reported bread/wheat flour consumption was positively associated with Hb. High-dose vitamin A supplementation during the previous 6 mo had an unexpected negative association with Hb. There was no significant change in mean ferritin from 2008/2009 to 2018/2019 (Ptrend ¼ 0.59) and a À2.79% unadjusted decrease in mean ferritin per year during 2013 to 2018/2019 (P-trend < 0.001). Fortifiable food consumption was not associated with ferritin. The consumption of MNPs in the previous 3 mo had an unexpected negative association with ferritin, though adherence was unknown.

Anemia and micronutrient deficiencies
Among WRA in Guatemala, there was a À0.02 g/dL unadjusted decrease in altitude-corrected mean Hb per year during 2008/2009 to 2018/2019 (P-trend < 0.001) and a À0.04 g/dL decrease in mean Hb per year during 2013 to 2018/2019 (Ptrend < 0.001). Table 4 shows potential factors associated with Hb and ferritin among WRA in Guatemala. Consuming standard sugar 3-7 d compared with 0 d/wk was positively associated with Hb. There was a À1.38% unadjusted decrease in mean ferritin concentration from 2008/2009 to 2018/2019 (P-trend < 0.001) and a À1.64% decrease in mean ferritin during 2013 to 2018/2019 (P-trend ¼ 0.002). Fortifiable food consumption was not associated with ferritin. There was a positive association between current use of reversible contraceptives, such that WRA users had on average and over the years had, higher levels of hemoglobin (þ0.21 g/dL) and ferritin (þ18.9 μg/L) relative to noncurrent users (each adjusted P < 0.001).
In adjusted fixed effects models of Hb and ferritin for PSC and WRA, there was no significant interaction between reported consumption of any fortifiable food and survey year indicating that the associations did not differ by survey year (data not shown).

Discussion
Guatemala has implemented multiple food fortification strategies with mandates on enrichment of wheat and nixtamalized maize flours with iron and other anemia-related micronutrients, and sugar fortification with vitamin A. Among PSC, Guatemala implements several supplementation programs, including iron, folic acid, and multiple micronutrient supplements, whereas in recent years, high-dose vitamin A supplementation has been limited to children aged 6-11 mo. The institutionalization of fortification and supplementation policies has coincided with declines in the prevalence of anemia among PSC and WRA over many years and the prevalence of iron deficiency anemia declined slightly among WRA. However, iron status did not improve during the 10-y period between 2008/2009 and 2018/ 2019 and worsened in 2017/2018. Fortification and micronutrient supplementation were not associated with improved ferritin status, and we were unable to identify clear timedependent and individual-level correlates of the recent increases in iron deficiency among PSC and WRA. The observed negative association between consumption of MNPs and ferritin was unexpected, and possibly due to confounding by indication of treatment [43] or enhanced MNP program implementation in areas of higher risks. However, the intensity of micronutrient supplementation interventions overall declined precipitously during the study period. Furthermore, young children aged 6 mo to 5 y are generally not expected to consume sufficient quantities of bread to be the primary beneficiary of the industrial fortification of wheat flour or maize flour, and previous studies have shown that wheat flour is not frequently consumed by the  [44]. Micronutrient supplementation and food fortification likely have more difficulty reaching rural and indigenous populations, which consume less industrially fortified flours and are often at a higher risk of iron deficiency [44]. The frequency of wheat flour consumption was positively associated with a higher Hb among PSC over the 2013-2018/ 2019 period in which there was a significant increasing trend in Hb concentrations. In contrast, there was a significant decreasing trend in Hb among WRA during the same period. Among WRA, reported standard sugar consumption was positively associated with Hb, potentially reflecting the role of vitamin A fortification in anemia prevention [9,45]. The plausibility of this pathway is strengthened by the strong association of RBP, a marker of vitamin A status, with Hb. Expected improvements in micronutrient status due to the introduction of maize flour fortification in 2016 may not have been observable during the survey because of typical delays in program maturation.
The iron fortification of bread met regulatory minimums for wheat flour fortification, and iron was present in fortification levels in one-fifth of sugar samples collected. Vitamin A fortification levels were also acceptable in over 80% of sugar samples. Current use of reversible contraceptives was positively associated with Hb and ferritin, and aligned with previous studies finding use of hormonal contraceptives as protective against anemia [46]. This study is strengthened by its data source, 6 nationally representative surveys over 10 y. These data sources are rich in variables pertaining to food fortification, supplementation, and micronutrient status, and used many comparable methods across time including the same laboratory, methodology, and many of the same well-trained personnel. These findings are likely generalizable to most of the 900,000 PSC and 4 million WRA in Guatemala. The variety of micronutrient indicators allows a more complete examination of the factors associated with anemia in the studied populations. This study also benefits from the use of food sample data to examine the level of fortification in key fortifiable foods available to the target populations. The use of the MPI is another strength as it captures many differing dimensions of poverty and aligns with Sustainable Development Goals and indicators [48]. Our analysis results are also consistent with external MPI estimates of the same period indicating that the SIVESNU nationally representative surveys adequately captured the demography of the country [15].
This study has several limitations. The ENMICRON survey, which occurred before the SIVESNU surveillance system was established, did not collect information on consumption of fortifiable foods, which meant this survey was dropped from relevant models. There were differences in the assessment of the biomarkers of nutritional status and inflammation between the 2008/2009 ENMICRON survey and the SIVESNU surveillance system, limiting their comparability. The blood source varied in 2013 and 2015 SIVESNU cycles (pooled capillary blood) compared with other years (venous blood). Food security could not be included in analyses due to the inconsistent use of tools across surveys. Food samples data could not be incorporated into the model because data were available for only a small subsample of households. The fortifiable food consumption frequency was evaluated over the previous 7 d and may not reflect usual consumption, though the consumption of fortifiable staple foods is less seasonally dependent, and the surveys were usually collected over 6-8 mo crossing multiple seasons. The food frequency data are also limited to foods prepared in the household and precludes potential interpretations related to street foods and those eaten outside. This is especially limiting to maize flour, which is most often consumed as premade purchased tortillas. Although the mandate for the fortification of nixtamalized maize flour was implemented during the study period, no maize flour samples were examined for micronutrient content. The surveys did not have dietary recall data or asked about the amounts of fortified food consumed, which is another limitation.
Guatemala has implemented multiple food fortification and micronutrient supplementation strategies to prevent several micronutrient deficiencies, including iron. The intensity of micronutrient supplementation interventions among PSC has declined between 2013 and 2018/2019. Although the country has made progress in the reduction of anemia among PSC, recent decreasing trends in ferritin among PSC and WRA, and Hb among WRA are concerning and may warrant renewed intervention in populations with higher risk and further monitoring. Secular trends toward poverty alleviation, education, and developmental might be responsible for changes not explained by the micronutrient interventions evaluated. Examining trends over time in both micronutrient program implementation quality and fidelity, in conjunction with micronutrient status assessment is useful for understanding program effectiveness, sustainability, safety, and accountability. These data are important for making evidence-based decisions to adapt, scale-up, or scale-down public health micronutrient delivering programs to improve their effectiveness and efficiency.

Author contributions
The authors' responsibilities were as follows -MP, KM: collected date and oversaw all technical aspects and training and were involved in the budgeting and acquisition of supplies/materials for the system. MP, DIM, CM, MEDJ, and RF-A: involved in the original design of the surveillance system. DIM, CM: provided technical assistance. MEDJ, RF-A, OYA, LG: provided inkind technical assistance. MCS, LG, OYA: were advisors to the surveillance system and all authors: read and approved the final manuscript.

Conflict of Interest
MP and KM were employed by the SIVESNU surveillance system for the data collection cycles reported in this manuscript and oversaw all technical aspects and training and were involved in the budgeting and acquisition of supplies/materials for the system. The other authors declare no other conflicts of interest.