Prevalence of anemia and deficiency of iron, folic acid, and zinc in children younger than 2 years of age who use the health services provided by the Mexican Social Security Institute

The first 2 years of life are crucial for children's present and future health and nutritional status and, more specifically, for their mental, physical, and emotional development. In the last decade, micronutrient deficiencies have received much attention as it has been demonstrated that even subclinical states, such as mild iron deficiency or low concentrations of zinc, are associated with functional outcomes [1]. These include, among others, impaired psychomotor development [2, 3], decreased work capacity, diminished immunological response, and linear growth retardation [4‐6]. Likewise, mild zinc deficiency has a negative influence on growth and development and increases the risk of diarrhea and acute respiratory infections [5, 7‐9]. Folic acid deficiency affects erythropoiesis, which, in turn, may be responsible for macrocytic anemia [10].

In Mexico, as in other developing countries, micronutrient deficiencies occur mainly between 6 and 24 months of age and are a significant public health problem [11‐15]. Children under 2 have increased nutritional requirements because of their growth spurt, which often leads to a negative nutrient balance [16, 17]. Preterm babies, as well as those born small for gestational age, are particularly vulnerable to iron deficiency in their first months [17]. During the first year of life, another risk factor for nutritional deficiencies occurs from inadequate complementary feeding practices. These are characterized by consumption of foods with low amounts of bioavailable micronutrients as well as with inhibitors of their absorption, and the practices often extend up to the age of 2 years. In the cases of iron and zinc, risk of inadequate intake is increased in countries where diets at this age include cereals with a high content of phytic acid coupled with low intake of animal foods [1, 7].

According to results from the National Nutrition Survey (NNS) in Mexico, carried out in 1999 (NNS-99), the highest prevalence of anemia was found in children between the ages of 12 and 24 months (48.9%); there was a slightly higher prevalence in rural areas (52.9%) as compared to urban ones (46.8%) [15]. In children under 5 years old, 35% of anemia cases were not associated with iron deficiency but rather with one or more vitamin deficiencies [13]. In children between 1 and 2, 66.6% of children had iron deficiency, as estimated by the percentage of transferrin saturation [13]. Folic acid deficiency was estimated in 8.8% of children under 2 and zinc deficiency in 34% [13, 14]. However, NNS-99 was not designed to be representative of children under 2 years old, so results for this age group are based on very small sample sizes.

When the study was conducted, the population of Mexico was approximately 100 million people. More than half of them were right-holders of the Instituto Mexicano del Seguro Social-IMSS (Mexican Institute for Social Security). Through two branches, IMSS provides social, welfare, and medical benefits for nongovernment employees and their direct dependents as well as for the underserved population living in rural or semi-rural areas. The regular regimen (RR) provides care to salaried employees, whereas IMSS-Oportunidades (IO) provides care to those in underserved, rural/semi-rural areas.

The present survey was carried out in primary health care-level facilities, which offer services for infants and children such as medical care for common childhood illnesses (i.e., the common cold, acute diarrhea) as well as well-baby clinic, including growth monitoring, promotion and support of breast-feeding, and primary preventive actions (i.e., vaccination). Additionally, several government-sponsored programs geared to the underserved population are offered through IO program like distribution of fortified foods for pregnant women, infants, and malnourished children. In communities where IO operated, the whole community is a beneficiary of this program and no other public medical services are available, making it unlikely that people living in those communities will use other services. Also, IO has a strong component of home visits and community development activities, which tend to strengthen the relationship between health providers and community members.

This study was commissioned by the IMSS to learn about the prevalence of anemia and micronutrient deficiencies, particularly of iron, folic acid, and zinc, in children under 2 years of age; it was part of a larger study to document the general health and nutritional status of this age group. To this end, we carried out a national survey, drawing two representative samples of children under 2 years of age, beneficiaries of RR and IO. For each of these populations, the sampling frame was designed to be representative of the four regions in which the country has been divided by other national surveys, which take into consideration differences in socioeconomic development (i.e., North, Center, South, and Mexico City (the nation's capital) with its surrounding periurban area, which includes the Federal District and some municipalities of the surrounding State of Mexico); IO does not operate in this last region, as no major semi-rural or indigenous population lives there. Within each of these regions, RR is based in urban areas, while IO provides its services in the rural/semi-rural areas. We expect that the results of this survey will provide the basis for public health interventions and program strengthening, oriented to alleviate and prevent micronutrient deficiencies in the target population.

The survey was carried out between 1999 and 2001 [18]. From this larger study, which included 35,997 children younger than 24 months of age, a subsample was selected to determine micronutrient status to assess the prevalence of iron, folic acid, zinc deficiency, and anemia. Regarding ethical considerations, the Institutional Research Review Board approved the protocol study and, the informed consent letter was signed by mothers of the children in the subsample. The larger sample followed a stratified, two-staged random model by which we obtained a representative sample for each of the two care regimens as well as for each of the four regions of the country that have been used by other national surveys in the past. The sampling frame from which primary sampling units were selected was drawn from a list of all primary health care units in the country, which included 1,160 family medicine units (FMU) for RR and 3,367 rural medical units (RMU) for IO. FMUs were organized by region and number of attending physicians in each of them; RMUs were organized by region and number of children under 2 living in the communities where these units were located. Secondary sampling units were drawn from the children younger than 2 years old who attended the out-patient clinic, vaccination clinic, and well-baby clinic, all based in the primary health care units already described.

Response rates were 64.2% (n = 2,646) in RR and 74.7% (n = 2,309) in IO. Response rates differed in different regions of the country. For RR in the North region, we were able to recruit 76.2% of the expected sample; in the Center, 73.9%; in the South, 58.7%; and in Mexico City and periurban area, 44.8%. For IO, the North recruited 85.8% of the expected sample; the Center, 93.6%; and the South, 44.8%. Children who participated in the micronutrient sample were significantly older than children who participated in the other parts of the larger study. In IO, there was also a significant difference in weight for age: children participating in the micronutrient sample had lower Z score (-0.604) than those in the larger sample (-0.318). The other characteristics–including sex, height for age, years of education of either parent, and percent of working mothers' or fathers' occupation–showed no statistically significant differences between the micronutrient and the general sample (Table 1). Between regions, the only significant difference was found in the years of schooling of RR of mothers coming from the North, which were less than in the rest of the sample.

At the time of this survey, 84% of children in RR and 94% of children in IO had attended the health care facility searching for some kind of preventive service, like a well-baby clinic or routine vaccination. Only 15.7% of children in RR and 6.4% in IO attended because of an acute illness. Additionally, 20% of mothers reported that the child had had one or more signs of acute infection (e.g., diarrhea, cough, nasal discharge, fever) during the two weeks before the study, which was an exclusion criterion for the micronutrient sample, as acute infections affect the serum status of zinc and iron, making interpretation difficult.

There were some sociodemographic differences in the study sample when comparing regimen by region. Parents' schooling was higher in RR as compared with IO; a larger percentage of mothers in RR worked outside home (24.4% vs. 8.2%); and the proportion of houses with dirt floors was lower in RR than in IO (4.3% vs. 33.1%). Other differences included: at 4 months of age, 23.8% of children in RR were exclusively breast-fed, in comparison with 46.2% in IO; at this age, children in the South had a higher prevalence of exclusive breast feeding compared to the rest of the country (31.5% in RR and 48.4% in IO); the proportion of children still receiving some breast milk at 12 months of age was larger in the South than in the rest of the country: 26.9% in RR and 66.2% in IO. Lastly, the prevalence of growth stunting in the country was 6.5% in RR, in contrast with 22.4% in IO, and growth stunting was more prevalent in the North for RR (7.4%) and in the Center for IO (24.2%).

In describing the sample included in this study, 13.2% of children in RR and 4.9% in IO were younger than 6 months of age; 38.2% and 31.0%, respectively, were between 6 and 11 months old; and 48.7% and 64.1%, respectively, were between 12 and 23 months of age. Sixty percent of children included in the sample had all four determinations for micronutrients, including Hb, serum ferritin, serum zinc, and folic acid. The most common reason why the other 40% did not have complete lab determinations was because of insufficient blood sample. Only 3% of the samples were lipemic or hemolyzed, and thus unsuitable for analysis. Because of budgetary constraints, only half the samples were analyzed in the North and Center regions of IO; samples selected for analysis were drawn at random.

Results in Tables 2, 3, 4, 5, 6, 7 are shown by region and age group in each branch of IMSS. Resorting to the age groupings used in other published health surveys in the country, age categories included: less than 6 months, 6–11 months, and 12–23 months.

The evidence presented in this article is disturbing in terms of public nutrition. It shows that children younger than 2 years old have several micronutrient deficiencies that may stop them from attaining their full growth and development potentials. One of every five children had anemia, and one of every three had iron deficiency. More than half the anemia cases were not accompanied by low ferritin levels, reflecting that they may be from causes other than iron deficiency. Low serum zinc concentrations were found in one out of three children living in the urban areas and in one out of ten of those living in rural areas; 10% of children of RR and 8% of children in IO had folic acid deficiencies. Low serum zinc and iron deficiency appeared together in 9% and 3% of children living in urban and rural areas, respectively.

Although we have the largest sample size published in the country in this age group, there are several possible sources of bias in our sample. The sample was only representative of children who use the health care services provided by the institution and who were free of acute infectious diseases during the two weeks before giving the blood sample. Although the decision to exclude data about children with acute infectious diseases ensured that the values of iron and zinc would not be affected by the infectious state, it is possible that children who were ill may have had a different micronutrient status than those included in the sample, introducing a further potential bias.

In RR, our study sample was only 64.2% of the calculated sample size; in IO, it was 74.7%; with differences by regions. In view of the different response rates, which led to underrepresentation of the South and Mexico City regions for RR and the South region for IO, there is a likelihood of biases in some of the results presented. The different regions in the country have noticeable contrasts in socioeconomic indicators, as the North is more developed than the South. Likewise, the two regimens care for populations with noticeable differences in economic development, favoring the urban areas over the rural/semi-rural. Other possible sources of bias emerge from the composition of the subsample of children who participated in the micronutrient study. In comparison with the rest of the children who participated in the larger study, those who were sampled for micronutrient status were older. Also, children in the IO subsample had lower weight for age Z scores. Therefore, it is possible that the true prevalence might be overestimated, as older children and those with lower weight for age may also be more likely to have micronutrient deficiencies. We also can not rule out the possibility that some characteristics of the study sample might have been influenced by parental acceptance of the study, which might have been affected by their own perceptions and concerns about their children's health status. It is difficult to estimate how the different sampling biases might have influenced the results, however, we are confident that, whatever the direction of the biases, the sampled population reflects the population who use the primary health care facilities of IMSS. To correctly interpret our results, it is necessary to consider some methodological issues. The criterion that we used to define iron deficiency was based only on ferritin values, defining a cutoff of <10 μg/L to reflect low iron stores. Other indicators of iron status, like the percentage of transferrin saturation (with a cut-off at <10%) were not useful, as it only classified 0.2% and 3.3% of children in RR and IO, respectively, as iron deficient. This may be because transferrin saturation has ample variation in the first year of life, so there is no agreement on the best cut-off [16, 30, 31]. In view of the low sensitivity that this indicator has to detect iron deficiency in this age group, we decided not to use it.

There is no general agreement about the preferred cut-off point for ferritin in this age group. We used a cut-off of < 10 μg/L, but other authors suggest a different one. Some authors have pointed out that the cut-off values for children under 5 should vary between 8 and 12 μg/L, depending on the specific age group [32‐36]. WHO [36] recommends using a cut-off of < 12 μg/L. We ran our analysis with this WHO-recommended cut-off and found a difference in prevalence of about 4 percentage points higher with it. In NHANES III (1988–1994), the value of the 5^th percentile for serum ferritin in children under 1 year of age was equivalent to 11 μg/L; the corresponding value for children between 1 and 2 years of age was 6 μg/L. However, for the Mexican American sample included in this study, the corresponding values were 8 and 3 μg/L, respectively [37]. Looking at healthy Honduran and Swedish children–exclusively breast-fed infants at 4 and 6 months of age–Domellöf et al. found -2 SD cut-off values corresponding at 4 mo of age to 20 μg/L, at 6 mo of age to 9 μg/L, and at 9 mo of age, to 5 μg/L [38]. Soh et al. have summarized this controversy, stating that until the validity of cut-off points of iron indices is confirmed, prevalence estimates of iron deficiency in children younger than 2 must remain conjectural [39].

Anemia, on the other hand, was defined based on hemoglobin values that were adjusted for each age category in the range of our study. The issue here is whether these cut-offs are adequate to identify anemia. The use of a single cut-off for children under 5 years of age has been a point for discussion [24]. Some authors consider that 11 g/dL, as proposed by WHO for this age group, may be too high for children less than 2 years old [16, 30, 34]. In fact, changes in iron metabolism found in this age, characterized by a rapid growth spurt and exposure to different infectious diseases, many of which happen subclinically, may affect the biochemical response of the iron status indicators, so there is uncertainty about the proper cut-offs that may best reflect functional outcomes [16, 30]. Sheriff et al. found that the 5th percentile for hemoglobin values for toddlers 8 months of age was 9.7 g/dL, whereas the corresponding value for children between 12 and 18 months was 10 g/dL [34]. Further, a study carried out in Sweden with healthy babies born at term found that more than 30% had hemoglobin concentrations under 11 g/dL by 6 months of age, although fewer than 3% had ferritin values less than 12 μg/L, so few had low iron stores [16].

In our own study, more than 50% of cases with hemoglobin under the age-specific cut-off were not associated with low ferritin values. It may be that, even when we used cut-off values adjusted for age, we overestimated the prevalence of anemia. The possibility exists that, although we left out any child with data of acute infection and our sample population was mostly drawn from the well-baby clinic and preventive programs, a certain proportion of cases might have had a subclinical infection that raised ferritin values; however, we do not think that this is the most likely explanation. Therefore, children who presented low hemoglobin and normal ferritin values were considered as having anemia from causes other than iron deficiency.

Other micronutrient deficiencies are also known to cause anemia; this is the case of folic acid deficiency, as well as others, like vitamin A and B₁₂, which were not assessed in our study [40]. Anemia without iron deficiency has been reported by other authors, in some instances in as many as over 50% of cases, a prevalence similar to the one found in our study [25, 41, 42]. It should be noted that this last figure also holds true when we analyzed our data using a cut-off of < 12 μg/L for serum ferritin to identify iron deficiency.

Even when most of the Mexican population is covered by the IMSS system, this population has several different characteristics. On the one hand, for RR, at least one of the parents has to have a regular salary to be an IMSS beneficiary. On the other hand, the populations served by IO are also beneficiaries of other subsidy and nutrition alleviation government programs, such as Oportunidades [43]. This and other programs are targeted at low-income households, and include the provision of micronutrient-fortified products for pregnant/lactating women and infants between the ages of 4 and 23 months old, as well as for underweight children aged 2–4 and may have some influence in improving the micronutrient status of these groups [44]. Thus, it is relevant to compare our data with those collected by the NNS conducted in 1999, in Mexico as both studies were carried out at approximately the same time. However, there were some methodological differences: our survey used venous blood for sampling, whereas NNS used capillary samples; our study used age-adjusted hemoglobin cut-offs values to identify anemia, whereas NNS used two cut-offs: 9.5 g/dL for children 6–11 months old (compared to 10.5 g/dL used in the present study for this age group), and 11 g/dL for children 12–23 months old (compared to 10.7 g/dL for this age group in our study) [13].

With these caveats in mind, we noted that children 6–11 months old in RR had a prevalence of anemia of 20.9 (95% CI = 17.0–25.4), higher than that found by NNS: 11.3% (95% CI = 11.2–21.0). In comparison, the prevalence found in children in IO was very similar to the one found in the NNS rural areas of the country: 15.5%, 95% CI = 11.2–21.0 vs. 16.2%, 95% CI = 11.8–20.6, respectively. In our study, children between 12 and 23 months old had lower prevalence of anemia than the general Mexican population reported by NNS, both in the RR and the IO areas: 22.7%, 95% CI = 19.2–26.6 vs. 46.8%, 95% CI = 43.1–50.5 in urban areas and 21.7%, 95% CI = 17.3–27.0 vs. 52.9%, 95% CI = 48.0–57.9, for the IMSS and the general population, respectively. Taking an external reference as comparison, in the 2002 report of the WHO for Latin America, the prevalence of anemia in children under 5 varied between 16% and 28% [45]. Although the prevalence that we found in our study lies well within this range, it is known that the prevalence of anemia is higher in the first 2 years of life. Therefore, the prevalence that we found in our population may be lower than that in other Latin American countries and was lower than that found in the general Mexican population.

Another interesting comparison was found when contrasting iron deficiency between RR and IO. In both regimens, iron deficiency without anemia and anemia were present since the first 6 months of life in 15% of children under RR and 32% of children under IO. Iron deficiency at this early age has often been found to be related to a combination of the poor iron stores accumulated over the last trimester of pregnancy and the poor quality of complementary foods, which may be either low in iron or in sources of highly bio-available iron [18, 39]. Further, it was clear that the prevalence of iron deficiency increased with age, a fact that calls attention to the need for early actions by primary health care public services.

With respect to other micronutrient deficiencies, the only statistical significant difference between RR and IO was a higher prevalence of low serum zinc concentration in the former. Although it would be expected to find higher prevalences of micronutrient deficiencies in the more underprivileged areas, the targeted food and nutritional assistance that the underserved population has received over the past years may be reflected in an improvement in their micronutrient status [44]. It is generally accepted that we do not have good indicators to assess zinc nutritional status, especially as related to biological responses that may translate functional outcomes of mild to moderate deficiency. Some studies have evaluated the adequacy of dietary consumption, whereas others have used plasma or serum concentration of zinc. The WHO has estimated the prevalence of zinc deficiency based on dietary intake, which has a mean of 31% worldwide, ranging between 6 and 73% in different populations in the under-5-year-old group [45]. The most widely used indicator, however, has been plasma or serum zinc concentration [7, 26]. Using this indicator, we found a high prevalence of zinc deficiency, particularly in children of RR. Even when the consequences of zinc deficiency have been fairly well documented in the literature and include limitations on growth potential, impaired psycho-motor development, impaired immune function, and delayed bone and sexual maturation onset [5‐7, 46, 47], the extent of zinc deficiency has not been properly established worldwide, nor are there specific public nutrition programs addressing it, aside from the recent zinc supplementation recommended for children with acute diarrhea [5, 46, 48].

Another result worth discussing is the presence of multiple micronutrient deficiencies. When compared to children with normal iron status, children with anemia had a higher prevalence of folic acid deficiency. Similarly, low plasma zinc concentrations were more prevalent in children with anemia. Although the prevalence of children with three simultaneous deficiencies of the micronutrients studied was relatively low, the coincidence of two of these deficiencies singled out those of zinc and iron as the most common deficiencies in RR children, and iron and folic acid in IO. These deficiencies will manifest themselves as anemia–not necessarily related to iron deficiency. They call for more comprehensive interventions–whether food based, supplement based, or fortification based–than the ones usually considered by public health programs. Further, the early age at which these deficiencies present themselves is a call for public health interventions to consider preventive approaches rather than curative ones.

Iron and zinc are the principal micronutrient deficiencies in Mexican children younger than 2 years old who use the health care services provided by IMSS. Anemia not associated with low ferritin values was more prevalent than iron-deficiency anemia. The fact that we found multiple micronutrient deficiencies at this early age reveals a need to establish effective preventive public health nutrition programs to address them.