|Home | About | Journals | Submit | Contact Us | Français|
Anemia affects almost two-thirds of pregnant women in developing countries and contributes to maternal morbidity and mortality and to low birthweight.
To determine the prevalence of anemia and the dietary and socioeconomic factors associated with anemia in pregnant women living in an urban community setting in Hyderabad, Pakistan.
This was a prospective, observational study of 1,369 pregnant women enrolled at 20 to 26 weeks of gestation and followed to 6 weeks postpartum. A blood sample was obtained at enrollment to determine hemoglobin levels. Information on nutritional knowledge, attitudes, and practice and dietary history regarding usual food intake before and during pregnancy were obtained by trained interviewers within 1 week of enrollment.
The prevalence of anemia (defined by the World Health Organization as hemoglobin < 11.0 g/dL) in these subjects was 90.5%; of these, 75.0% had mild anemia (hemoglobin from 9.0 to 10.9 g/dL) and 14.8% had moderate anemia (hemoglobin from 7.0 to 8.9 g/dL). Only 0.7% were severely anemic (hemoglobin < 7.0 g/dL). Nonanemic women were significantly taller, weighed more, and had a higher body mass index. Multivariate analysis after adjustment for education, pregnancy history, iron supplementation, and height showed that drinking more than three cups of tea per day before pregnancy (adjusted prevalence odds ratio [aPOR], 3.2; 95% confidence interval [CI], 1.3 to 8.0), consumption of clay or dirt during pregnancy (aPOR, 3.7; 95% CI, 1.1 to 12.3), and never consuming eggs or consuming eggs less than twice a week during pregnancy (aPOR, 1.7; 95% CI, 1.1 to 2.5) were significantly associated with anemia. Consumption of red meat less than twice a week prior to pregnancy was marginally associated with anemia (aPOR, 1.2; 95% CI, 0.8 to 1.8) but was significantly associated with lower mean hemoglobin concentrations (9.9 vs. 10.0 g/dL, p = .05) during the study period. A subanalysis excluding women with mild anemia found similar associations to those of the main model, albeit even stronger.
A high percentage of women at 20 to 26 weeks of pregnancy had mild to moderate anemia. Pica, tea consumption, and low intake of eggs and red meat were associated with anemia. Women of childbearing age should be provided nutritional education regarding food sources of iron, especially prior to becoming pregnant, and taught how food choices can either enhance or interfere with iron absorption.
Anemia is one of the most common nutritional deficiency diseases observed globally. Although nutritional anemia affects members of both sexes and all age groups, the problem is more prevalent among women and contributes to maternal morbidity and mortality, as well as to low birthweight . It has been estimated that nutritional anemia affects almost two-thirds of pregnant women in developing countries. However, many of these women were already anemic at the time of conception, with an estimated prevalence of anemia of almost 50% among nonpregnant women in developing countries . In Pakistan, the prevalence of anemia among ever-married women aged 15 to 44 is reported to be 26% in urban areas and 47% in rural areas . The prevalence of anemia among pregnant women living in urban areas is similar, ranging from 29%  to 50% among pregnant women attending antenatal clinics in a large private, tertiary hospital in Karachi [4, 5].
The etiology of anemia during pregnancy among women in developing countries is multifactorial and varies by geographic region . The primary cause of anemia during pregnancy worldwide is iron deficiency secondary to chronic inadequate dietary intake and menstruation, heightened by the physiologic demands of the fetus and maternal blood volume expansion during pregnancy [6, 7]. Genetic causes and poor hygiene that may lead to infections and infestations are other contributing factors .
With limited resources available to address public health problems, knowledge of the local etiological factors responsible for anemia is crucial in order to design appropriate prevention and treatment strategies. Most of the published studies from Pakistan have been conducted on women seeking care in clinical or hospital settings and thus may not give a true picture of anemia and its causes in a population-based sample. The aim of our study was to report the prevalence of anemia and the dietary and socioeconomic factors associated with anemia in a cohort of pregnant women living in an urban community setting in a developing country.
This is a cross-sectional analytic study, nested in a prospective population-based cohort study conducted in Pakistan to assess the association of various maternal characteristics with pregnancy outcome. The study was part of a research collaboration to improve maternal and child survival in Pakistan involving the University of Alabama, Birmingham, Alabama, USA; the Research Triangle Institute, Durham, North Carolina, USA; and Aga Khan University, Karachi, Pakistan. The institutional review boards of all three institutions approved the study, and all subjects provided informed consent.
The project was conducted from 2003 to 2005 in the city of Hyderabad, which is situated in Pakistan’s Sindh Province and has an estimated population of 1.1 million people . All pregnant women in the catchment area were screened by government-trained Lady Health Workers for eligibility. Those eligible to participate were invited to visit the project’s Hyderabad research clinic for further assessment. At the clinic, women who met the inclusion criteria (at 20 to 26 weeks of gestation, permanent residents of the study area, and consenting to the study) were recruited into the study .
Participants were interviewed at the study clinic regarding demographic and socioeconomic characteristics and pregnancy history. Physical examination was performed by the study physicians or midwives, and blood samples were collected. The body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in meters. Nutritional assessment with the use of a validated instrument from Pakistan was completed in the homes of the participants within a week of enrollment .
Information on knowledge, attitudes, and practices regarding food intake during pregnancy and dietary history of food intake before and during the current pregnancy was obtained with the use of a structured questionnaire by trained personnel. In addition, a 24-hour dietary recall was obtained .
At the time of enrollment, 3 mL of venous blood was collected by a trained technician using standard procedures. The hemoglobin concentration was determined by Sahli’s method . Anemia was classified according to the World Health Organization (WHO) classification for pregnant women . Mild anemia was classified as hemoglobin concentrations of 9.0 to 10.9 g/dL, moderate anemia as hemoglobin concentrations of 7.0 to 8.9 g/dL, and severe anemia as hemoglobin concentrations < 7.0 g/dL. A study participant was considered nonanemic if her hemoglobin concentration was ≥ 11 g/dL.
Data entry in Microsoft Access included range and validation checks. Descriptive analysis and regression models were run in SPSS for Windows, version 13. Descriptive data are presented as prevalence rates of anemia and mean hemoglobin concentrations. Means and percentages were compared by Student’s t-test, analysis of variance, or the χ2 test, as appropriate. The probability of being anemic was evaluated by a multivariate logistic regression model. Variables that did not meet the p value criterion but had known biological or social significance were also considered in development of the model. The independent variables with a moderately low p value (p < .25) or those that had been documented previously to be associated with anemia were included in the model: socioeconomic status, maternal education, employment status, pregnancy history, pica, and dietary intake before and during the current pregnancy.
Correlations among the various independent variables were evaluated. The modeling exercise began with the variable found most significant in the univariate analysis and continued with the subsequent addition of the next most significant variable. The final model included variables that were significant at p ≤ .10.
A subanalysis was also done by excluding women who were mildly anemic from the dataset in order to see whether similar associations as those in the full model would be observed based on only moderate and severe cases.
The final sample cohort included 1,366 women. Three were excluded because of missing dietary data. The age distribution and the socioeconomic characteristics of the excluded cases were not different from those of the final sample. The prevalence of mild to moderate anemia (hemoglobin from 7.0 to 10.9 g/dL) in the cohort was 89.8%; of these, 75.0% had mild anemia (hemoglobin from 9.0 to 10.9 g/dL) and 14.8% had moderate anemia (hemoglobin from 7.0 to 8.9 g/dL). Only 0.7% were severely anemic (hemoglobin < 7.0 g/dL) (table 1).
Table 2 compares anthropometric and demographic characteristics of moderately to severely anemic, mildly anemic, and nonanemic women. Moderate to severely anemic subjects were slightly older than nonanemic subjects; whereas nonanemic women were significantly taller and heavier, and a lower proportion were underweight (BMI < 18.5). In addition, anemic women were more likely to have no formal education and to be employed outside the home. These differences were significant when nonanemic women were compared with mildly anemic women or with moderately to severely anemic women, except for formal education status, for which no differences existed between nonanemic women and mildly anemic women. The only ethnic group that had a significantly lower proportion of anemia was the Pushto-speaking women. The prevalence of moderate to severe anemia among Pushto-speaking women was the lowest (2%) among the ethnic groups (p < .001).
We also compared mean hemoglobin concentrations in relation to various maternal characteristics. The number of prior pregnancies was inversely related to mean hemoglobin level (p = .018). Women reporting a prior stillbirth also had lower mean hemoglobin levels than those who did not (9.69 vs. 9.97 g/dL, p = .0076) (table 3).
The mean hemoglobin concentration was significantly lower among women with no formal education than among women with some formal education (9.7 vs. 10.0 g/dL, p = .015), but there was no significant difference in hemoglobin levels between employed and unemployed women (9.7 vs. 9.9 g/dL, p = .65).
Table 4 presents the dietary characteristics associated with anemia levels and the mean hemoglobin concentrations. Women who reported consumption of red meat or chicken two or more times per week before pregnancy had higher hemoglobin concentrations, but only the differences in mean hemoglobin concentrations associated with consumption of red meat were significant (10.03 vs. 9.87 g/dL, p = .004). Increased tea consumption before pregnancy was associated with a significantly higher prevalence of anemia, but the differences in the mean hemoglobin concentrations were only of borderline significance (p = .065).
During pregnancy, reported dietary consumption of meat and tea was also associated with less anemia and, as in prepregnancy, with a significant increase in mean hemoglobin concentrations in those who consumed greater amounts of red meat. Also, as was the case for tea consumption prior to pregnancy, the greater the quantity of tea consumed during pregnancy, the lower the mean hemoglobin concentration. Consumption of fruits and eggs more than twice a week during pregnancy was associated with significantly higher hemoglobin levels.
Consumption of nonfood items, especially clay and dirt were strongly associated with anemia, as well as with lower mean hemoglobin concentrations. Intake of iron supplements was associated with an increasing prevalence of anemia and lower hemoglobin concentrations.
In the final multivariate model (table 5), tea consumption before pregnancy (adjusted prevalence odds ratio [aPOR], 3.2; 95% confidence interval [CI], 1.3 to 8.0), egg consumption less than twice a week or never during pregnancy (aPOR, 1.7; 95% CI, 1.1 to 2.5), and consumption of clay or dirt (aPOR, 3.7; 95% CI, 1.1 to 12.3) were associated with anemia after adjustment for education, pregnancy history, iron supplementation, and height. Consumption of red meat less than twice a week prior to pregnancy was marginally associated with anemia (aPOR, 1.2; 95% CI, 0.8 to 1.8).
A subanalysis was conducted on 341 women (130 nonanemic women and 211 moderate to severely anemic women); 1,024 women with mild anemia were excluded. The purpose of the subanalysis was to determine if the associations seen in the main model remained when mildly anemic women were excluded. In the univariate subanalysis, consumption of meat less than twice a week or never before pregnancy was associated with an increased risk of moderate to severe anemia. The consumption of tea more than three times per day before pregnancy was strongly associated with an increased risk of anemia. Increased tea consumption during pregnancy was also associated with moderate to severe anemia, although the increase was not statistically significant. Consumption of clay and dirt had the strongest association with anemia during pregnancy. In the final multivariate model of the subanalysis, all the associations were similar to those in the main model, albeit even stronger.
The prevalence of anemia in our study population (90.5%) was higher than that reported by other studies from Pakistan. One possible explanation for the high rate of anemia could be that the hemoglobin measurements were obtained late in the second trimester, when the hemoglobin concentration is expected to be at its lowest because plasma volume is expanding more rapidly than the red cell mass. This is consistent with the observation that 75% of the cases of anemia were mild. A similar high rate of anemia was found in a large Indian study including 11 states, where the prevalence of anemia was 87% among 4,775 pregnant women at more than 20 weeks of gestation . The prevalence of anemia among pregnant women in other developing countries ranges from 35% to 81% [1, 8, 13-15].
However, the prevalence of severe anemia was lower in our study than the 2% to 7% reported for developing countries . Nearly half of the women had borderline anemia, with hemoglobin concentrations more than 10 g/dL but less than 11 g/dL. These are encouraging findings for policy makers, because the effort required to reduce the prevalence of anemia in nearly half of pregnant urban women should not be as great as that needed to produce normal hemoglobin levels in women with more severe anemia. Prepregnancy screening with appropriate nutritional counseling or iron and vitamin supplementation may be the best intervention to reduce the high prevalence of pregnancy-related anemia. On the other hand, mild anemia may be due to other causes, such as thalassemia minor, which would have very different policy implications.
In our study, anemic women had a lower BMI than nonanemic women. However, the mean midpregnancy BMI of severely anemic women was 22.1 ± 3.7, suggesting that these women had adequate energy intake but that their diets may have been deficient in dietary iron or that they had poor iron absorption. The anemic women in our study were also significantly shorter than the nonanemic women, suggesting a sustained pattern of undernutrition in their early childhood and implying not only inadequate caloric or nutrient intake but also an inadequate overall diet.
We found the hemoglobin concentration of employed women to be lower that that of unemployed women. This was not an unexpected finding in our study setting, where women often need to work outside the home because of low family income. Furthermore, most of these women would be considered underemployed, i.e., employed but with jobs that were unsteady and low-paying, as suggested by comparison of the median monthly income of anemic women (Rs 750) with that of nonanemic women (Rs 857). However, the impact of employment on anemia status should be interpreted with caution, since the number of employed women was relatively small.
Interestingly, the prevalence of anemia was above 90% in all ethnic groups except the Pushto-speaking women. This is consistent with the results of both the 1987 and the 2001–02 National Nutrition Surveys [16, 17], which found the prevalence of anemia among pregnant and lactating women in the North West Frontier Province to be 6.9% and 31.8%, respectively, as compared with the national average of 45.5%. A possible reason for the lower prevalence of anemia in the Pushto-speaking women could be differences in their dietary practices. In general, diets in most Pakistani populations are heavily reliant on whole wheat flour, which is high in phytate that is known to interfere with iron absorption. On the other hand, the traditional staple of the Pushtoon is naan, which is made from refined flour that undergoes fermentation. Refined flour is lower in phytate, and the process of fermentation breaks down the phytate that may be present in the flour, thus resulting in improved iron absorption [18, 19]. We were unable to conduct a dietary intake subanalysis according to ethnicity because of an insufficient sample size.
In univariate analysis, consumption of fruit two or more times per week was associated with a decreased risk of anemia. Given the fact that a large percentage of the iron in these diets is from nonheme sources, the decreased risk may be attributed to the presence of vitamin C, which is known to enhance the absorption of nonheme iron .
Women who reported eating meat two or more times a week had higher mean hemoglobin concentrations. This was an expected result, since meat is an important source of heme. However, in the final adjusted model, in the presence of dietary intake of meat prior to pregnancy, dietary intake of meat during pregnancy was no longer significantly associated with anemia status. Although covariance may explain this finding, it may also indicate that the length of pregnancy is too short for an improved diet to have a significant impact on nutritional status among women who are undernourished before conceiving. Iron supplementation alone during pregnancy may be inadequate to prevent anemia in a large proportion of women who enter pregnancy with little or no iron stores. Furthermore, reported compliance rates for iron supplements during pregnancy are low [21-23]. Therefore, if pregnancy is viewed as part of the reproductive cycle, the appropriate time for intervention is before pregnancy rather than during pregnancy . WHO recommends that in vulnerable populations, supplemental iron should be given to adolescents and women for 2 to 4 months per year to ensure that women have reasonable iron stores when they enter pregnancy . Furthermore, current nutritional recommendations for improving the outcome of pregnancy emphasize the importance of ensuring that women are in good nutritional status prior to conception; hence, the focus needs to be shifted from diet during pregnancy to diet for the childbearing years, especially in developing countries where adolescent pregnancies are common and access to health care may be limited .
In terms of dietary practices, it is interesting to note that the consumption of red meat more than twice weekly tended to be protective, and tea consumption had a negative effect. Given the socioeconomic status of the majority of Pakistani women, it may not be possible to increase the intake of heme iron substantially, but nutrition education to include red meat even as part of a mixed vegetable or legume dish could reduce the prevalence of anemia by not only increasing the intake of iron but also enhancing the intake of nonheme iron, especially from wheat flour, which is the staple food in this area. Second, all women whose intake of iron is not optimal should be educated regarding the detrimental effect of tea on iron absorption  and should be counseled to decrease tea intake or at least space the intake of tea away from mealtimes. In addition, pregnant women should be counseled regarding the detrimental effect of pica, especially the intake of clay and dirt, on iron absorption [27, 28].
Contrary to expectation, women who reported consuming iron supplements had a significantly lower hemoglobin concentration. It is possible that only those women who were identified as being anemic or had symptoms of anemia were provided or reported taking iron supplements. Finally, we found low socioeconomic status, poor diets, increasing gravidity, and low educational status to be associated with anemia, as has been reported previously in Pakistan .
We were unable to account for other known factors associated with anemia, such as helminth infection, genetic disorders, or other nutritional deficiencies such as folate, vitamin B12, and vitamin A deficiency. Although helminth infection contributes to the prevalence of anemia , it may not be an important factor in this area because of its low prevalence in Pakistan .
Factors in our population that may merit exploring are genetic disorders such as thalassemia. Carriers of beta-thalassemia minor usually either are asymptomatic or present with mild anemia, and in the presence of numerous other potential contributing factors to anemia, beta-thalassemia minor may be overlooked.
Although iron stores can be mobilized to meet the increased requirements of pregnancy, the high prevalence of anemia in these pregnant women strongly suggests that the women generally had low stores of iron and probably had preexisting anemia. Thus, even the compensatory increase in iron absorption during pregnancy may be insufficient to prevent anemia in these women. We found that a high percentage of women in their third trimester of pregnancy had mild to moderate anemia and that dietary intake before as well as during pregnancy was associated with anemia. Therefore, the role of adequate diet and optimal prepregnancy nutritional status needs to be emphasized in women of childbearing age, particularly since pregnancy may be too short a period of time in which to reduce preexisting anemia. These recommendations would hold true in developing countries such as Pakistan where poor women generally do not seek prenatal care and those who do may not do so until the third trimester.
The causes of anemia need further research; in particular, a focus on less commonly studied factors may be required to develop adequate interventions. However, in the meantime, it is recommended that all women of childbearing age in Pakistan should receive basic nutrition education regarding food sources of iron and how food choices affect iron absorption.
This work was funded through grants from the National Institute of Child Health and Human Development (NICHD) HD40607-04 and HD40636-06 and the Bill and Melinda Gates Foundation for the Global Network for Women’s and Children’s Health Research (NICHD/NIH, USA). The authors wish to thank the data collection and data management team of the Aga Khan University Community Health Sciences Department, Pakistan.
Naila Baig-Ansari and Salma Halai Badruddin were responsible for the overall design of the nutrition component, analysis of the data, and writing the paper. All authors were involved in the design or implementation of the study and reviewed the paper. The authors have no conflicts of interest relevant to this study.