Socio‐economic and dietary diversity characteristics are associated with anemia among pregnant women attending antenatal care services in public health centers of Kembata Tembaro Zone, Southern Ethiopia

Abstract Anemia affects every country in the world including Ethiopia. It costs the socio‐economic developments due to its health impacts. This study was aimed to determine socio‐demographic and dietary diversity and other factors associated with anemia among pregnant women in Kembata Tembaro Zone. Institution‐based cross‐sectional study design was employed. Data were collected from 423 pregnant women. Hemoglobin level was measured by Hemocue. Pretested interviewer‐administered close‐ended questionnaire was used to collect the data. Analysis was done using SPSS for windows version 20. Descriptive and multivariate logistic regression analysis was conducted. We found that overall prevalence of anemia was 18%. Less than half (43.3%) of the respondents visited the health centers for antenatal care follow‐up at least once during their pregnancy time. Only 20.1% consumed more than five food preceding 24 hr. 59.3% of the respondents were taking iron supplements during data collection time. Anemia was significantly associated with the lowest wealth index, formal education, women's occupation, husband occupation, low dietary diversity, and inadequate intake of iron‐rich foods, history of malaria infection, maternal age, and parity. Anemia was of mild public health concern in the study area. The uptake of antenatal care service for the recommended number of visits and iron folic acid supplementation and the diet diversity was low in the pregnant women. The prevalence of anemia was mild public health concern. The early and timely uptake of iron folic acid supplementation was poor. Dietary diversity and consumption of iron‐rich foods was found to be low. Thus, community mobilization and health education to improve the uptake of the antenatal care services, increase micronutrient intake through food based and control of infections are recommended.


| INTRODUC TI ON
Anemia affects every country in the world. Basically, it costs the socio-economic developments due to its health impacts (Benoist, McLean, Egli, & Cogswell, 2008). There are various factors which contribute to the onset of anemia. From the major factors, low level of iron nutrient intake and infections like malaria play an important role for the occurrence of anemia. The main function of the iron as a nutrient is carrying oxygen, and its low level in the diet mainly which is manifested in the large proportion of humans across the globe is the main factor for it (Zimmermann, 2007), but also it often works in symphony with folate and/or vitamin B 12 deficiency (Milman, 2014).
These micronutrient deficiency anemias, particularly of iron combined with infections such as malaria, are common in Ethiopia (Ezechi Oliver, 2012). Worldwide, it is estimated that about 20% of maternal deaths are caused by anemia; in addition, anemia contributes partly to 50% of all maternal deaths. Similar situation is found in sub-Saharan Africa where anemia is reportedly accounted for about 20% of all maternal deaths (Oladunjoye, 2012). Anemia among the pregnant women has both maternal and fetal impacts (Milman, 2014 (19%), and women who are neither pregnant nor breastfeeding (15%). The prevalence also differs by residence areas (urban vs. rural). Greater proportion of women in rural areas were anemic (18%) compared with the urban areas (11%). From the other socio-demographic characteristics, educational status was a factor for anemia. Thus, women with no formal education got double probability of having anemia compared with the women who attended secondary school and above school. Correspondingly, the prevalence was decreased as the wealth status of the household becomes better (CSA, 2011). Irrespective of the national-level study, careful examination of the health institution-based reports exposed that anemia has been a serious health issue among the pregnant women in the study area. Moreover, the staple diets in the community are plant-based foods known to reduce iron absorption. Malaria and intestinal parasites are also in the top leading causes of morbidity; however, the most contributing factor for anemia is not well known specifically in Kembata Tembaro Zone, Ethiopia. Thus, this study is aimed to assess the prevalence of anemia and factors associated with anemia among pregnant women attending antenatal care service in public health centers.

| Study area
The study was carried out from August to September 2014 in public health centers of Kembata Tembaro Zone, Southern Ethiopia. The zone is one of the 15 zones in Southern Nations and Nationalities People Region (SNNPR). The capital town of the zone, Durame, is located 306 km from Addis Ababa through Hosanna, in the southwest direction. The town has altitude about 2,100 m above sea level (masl). The zone is divided into seven districts, one town administration, 120 rural kebeles and 22 urban kebeles/localities,

| Study design and population
The study implemented institution-based cross-sectional study design. The study participants were pregnant women who were selected during their visit in the selected public health centers in the study area for antenatal care. Those pregnant women who fulfill the inclusion criteria were enrolled in the study regardless of their number of visits. Each participants were recruited to the study only once on their all visits during the study period.

| Sample size and sampling technique
A total of 436 pregnant women were selected for the study. Single population proportion formula shown below was used with the assumption of 95% confidence level (1.96), margin of error of (0.05), prevalence of (22%) (CSA, 2011), and design effect of 1.5.
where n = required sample size. Zα/ 2 = critical value for normal distribution at 95% confidence level which equals to 1.96. d = an absolute precision (margin of error 5%).
Regarding sampling technique, of the 15 zones and four special district of SNNPR, Kembata Tembaro Zone was selected purposively.
The zone contains 28 public health centers; ten of the health centers were selected randomly. Then, the study units were assigned for each health center using proportional allocation method based on their catchment population. Lastly, systematic sampling method was used to include pregnant mothers who fulfill the inclusion criteria.
The list of pregnant women found at each health center and the last three-month average report review were used to estimate the number of pregnant women that would visit the health centers for antenatal care (ANC). The first respondent was selected by lottery method, and K = N/n (where N, average 1-month report and n = allocated sample size for each health center) interval was used to select the next respondent.

| Variables
Anemia was considered as dependent variable. The independent variables were socio-demographic variables and economic factors, maternal-related factors, and health-and nutrition-related characteristics.

| Blood sample collection and laboratory analysis
Capillary blood samples were used to assess hemoglobin concentration using HemoCue ® portable system, and the Hgb concentration was determined instantly after sample collection. Altitude was measured at the health centers using Magilan ® Global Positioning System (GPS). The hemoglobin concentration was adjusted for altitude according to WHO recommendation (WHO, 2011a, 2011b).

| Quality control
Data were collected by trained diploma and bachelor degree nurses.
One data collector for each health facility and four supervisors were recruited to overlook the entire data collection. In addition, one laboratory technician for each facility was recruited for blood sample collection and hemoglobin determination. The interview tools were adapted according to the existing scientific literatures and contextualized as per the locality and objectives of the current research. The interview tools were first prepared in English language and translated to the local language (Kembatigna) and back-translated to English by the native speakers of the languages. Later, two-day training was given for data interviewer and supervisors on the data collection tools. The consistency of the data was checked by the principal investigator and supervisors in daily basis. The accuracy of HemoCue ® had been checked daily by the use of control cuvettes provided with the machines.

| Data analysis
The data were entered and cleaned using Epi_info software version 3.5.1, and the analysis was done using statistical package for social sciences ( Principal component analysis (PCA) was conducted to construct the wealth index. The main variables included in the analysis were related to ownership of selected household assets, size of agricultural land, quantity of livestock, materials used for housing construction, and ownership of improved sanitation facilities. Lastly, the scores were divided into five equal wealth quintiles (lowest, lower, middle, higher, and highest).
The dietary diversity score (DDS) was computed from the nine food groups. Consuming a food item from any of the aforementioned groups earned a score of 1 for that specific category. If not, a score of 0 was given. Accordingly, a score of nine score points was developed.

| Socio-demographic characteristics of the respondents
Of the 436 respondents recruited, 423 (97%) volunteered to participate in the study. The mean age of the respondents was 26.85 with standard deviations of ±5.3 years, while their ages ranged from 18 to 43 years. Of the respondents, 256 (60.5%) were from rural areas and the rest 167 (39.5%) were urban dwellers. Almost all (98.8%) of the study participants were married at the time of data collection (Table 1).

| Prevalence of anemia from the laboratory findings among the pregnant women
Prevalence of anemia (Hgb <11 g/dl, adjusted for altitude) among the pregnant women was found to be 18%. Meanwhile, the mean (SD) Hgb level was 12.26 (±0.07) g/dl (95% CI: 12.13-12.41 g/ dl). The level for each trimesters was as follows 12.85 (±1.43) g/ dl (first trimester), 12.35 (±1.6) g/dl (second trimester), and 12.27 (±1.6) g/dl (third trimester). Again, the prevalence for the women who do not know their gestational age was 11.7(±1.4) g/dl. The occurrence of mild and moderate anemia was 37 (8.8%) and 39 (9.2%), respectively ( Figure 1). Besides the laboratory finding of hemoglobin level to diagnose anemia was supported by clinical findings (history and physical examination). Of the total 45 pale cases, 35 (77.8%) had hemoglobin concentration <11.0 g/dl. Those with previous history of anemia and those who had at least one symptom of anemia during their current pregnancy became anemic than their counterparts.

| Obstetric characteristics among the pregnant women
Majority, 260 (61.5%) of the respondents were in their third trimester, and 101 (23.9%) of them were in their second trimester, and 24(5.7%) were in their first trimester. The rest did not know

| Dietary diversity and other related characteristics of the pregnant women
Of the nine food groups, starchy staples (cereals, roots, and tubers)

| Socio-demographic and other factors associated with anemia
In the current study, it is revealed that wealth index, respondents educational status, women's occupation, husband's occupation, age of the women, and their residence variables were significantly associated with occurrence of anemia. Respondents from the lowest wealth quintiles had higher risk of developing anemia in contrast to highest wealth quintiles (AOR = 5.03, 95% CI = 1.07-23.7). Pregnant women with no formal education were the most affected compared with their counterparts (AOR = 6.26, 95%CI = 3.11-12.57). Daily laborers were more affected (AOR = 3.17, 95%CI = 1.1-9.14) than housewives. The pregnant women who were daily laborers were also highly affected by anemia (AOR = 3.1, 95%CI = 1.25-7.64) compared to farmer occupation. Moreover, the pregnant women residing in the rural did get highest odds of having anemia (AOR = 2.43, 95%CI = 1.08-5.44 (Table 3).
Beyond socio-demographic factors, dietary, obstetric, and other healthcare-related correlates were also associated with anemia. The likelihood of having anemia among the respondents with lowest dietary diversity (DDS ≤3) was found to be 3.66 times with 95%CI = 1.71-7.84 compared with higher dietary score (DDS ≥6. Similarly, respondents who does not eat iron-rich foods in the previous day of data collection had the odds of 4.3, 95% CI = 1.75-10.53). Dietary diversity was found to be better with the highest wealth quintiles of socio-economic status, urban residence, and higher educational level (Figures 2 and 3).
History of malaria attack was significantly associated with increased anemia (AOR = 7.86, 95% CI = 2.67-23.11). Pregnant women who gave birth at least once were more affected than pregnant women with the first pregnancy (AOR = 3.9, 95%CI = 1.57-9.69). In contrast to pregnant who received iron folic acid tablets, the odds of anemia was (AOR = 1.7, 95%CI = 0.91-3.24) higher elevated than women who did not receive the tablets, though the association was not significant (Table 3). The hemoglobin levels and prevalence of anemia were not statistically different across variables of gestational age, number of antenatal care visits, iron folic acid intake, frequency of coffee intake, place of birth, and birth interval in multiple variable logistic regressions.

| D ISCUSS I ON
The prevalence of anemia among pregnant women in the present study was 18%, 95%CI = 14.4-21.1 indicating that anemia in pregnant women was of mild public heath significance in the study area as per the World Health Organization (Benoist et al., 2008). The prevalence was closer but lower than the national prevalence (22% in 2011 and 24% in 2016) but higher than the regional prevalence (11%) reported in the EDHS (2011) (CSA, 2011(CSA, , 2016. It is also comparable to the findings from Addis Ababa (21.3%) (Hailu, 2014) and Gonder (21.6%) (Alem et al., 2013) but it is higher than the findings reported from Debre Berhan (9.7%) (Ayenew, Abere, & Timerga, 2014). The possible reason for the difference might be the methodology used to collect the data.
Unlike the current study, the Debre Berhan's study used a convenient method to include respondents into the study, lower sample size, and the hemoglobin findings were not adjusted for altitude. The finding of this study was however very low compared to a report from India, 94.3% (Kendre, 2014). It is suggested that the variations may be attributed to the different causes of anemia and dietary differences, population differences, and difference in methodology used in determining hemoglobin levels (Sahlis method vs. Hemocue method).
This finding correlates with the study done in Gonder, in which mild anemia 10.7%, moderate anemia (9.9%), and severe anemia (1%) were reported (Alem et al., 2013). The wealth quintiles of the and 1.83 (95% CI: 1.13-2.96), respectively, than secondary and above level of education (Gebremedhin, 2014). A study in India also shows prevalence of anemia decreases with increased educational level of women. The prevalence of anemia was as high as 88.2% among illiterates while it was only 34.1% in those educated up to intermediate level (Swarnlatha, 2013). In this regard, educational level may improve awareness for good nutrition practices during pregnancy and during their course of life. Thus, the improved nutrition reduces the vulnerability of anemia among the pregnant women. Maternal and husband occupations were related with increased risk of anemia in the current study. Odds of being anemic increases by 3.2 among daily laborers compared with housewife. Wives of daily laborers were affected three times, compared to wives of farmers.
This finding is consistent with finding in Nepal, where severe anemia was significantly more prevalent among women worked for wages (Zeina Makhoul, 2012). It is obvious that the occupation of individuals is affected by their educational level and the place where they live. Most of the time those who are engaged in different daily labor and small business activities are poorly educated and the most economically disadvantaged categories, which in turn affect their practice toward healthy and nutritionally adequate foods. Foods rich in iron like beef, chicken, and fish are very expensive that those low income, the daily wage laborers cannot purchase them. Moreover, they miss their meals since they are extremely busy throughout the day.
The occurrence of anemia among the pregnant women residing in the rural was higher than in the urban area. Rural residents were 2.4 times more affected by anemia than the urban dwellers. Similar result was reported in Ethiopia with AOR of 1.99 (95% CI: 1.73-2.30) (Tadege, 2009). In another study from southeast Ethiopia, residence in rural area affected (AOR = 3.3, 95% CI: 1.5-7.4) compared to those in urban area (Kefiyalew, Zemene, Asres, & Gedefaw, 2014). The higher prevalence of anemia among pregnant women from rural areas is likely related to lack of information about adequate nutrition during pregnancy, economic factors, and inaccessibility of healthcare centers. Anemia was significantly associated with history of malaria attack, and the association was retained even when the variable was adjusted for its confounding effect in the logistic regression models. Similar findings were observed in a study done in Benin, 15% (95% CI = 13-17) of anemia in this population was attributable to malaria infestation (Oue´Draogo, 2012). A study among pregnant women in Gilgel Gibe dam area, Southwest Ethiopia also found that the odds of anemia was higher among pregnant women who were malaria infected (AOR = 11.19, 95% C.I: 3.31-7.7 (Getachew, Yewhalaw, Tafess, Getachew, & Zeynudin, 2012). It is a known fact that malaria is one of the factors which cause and aggravate the occurrence of anemia in pregnancy. This is because of hemolysis or breakdown of red blood cells, affected by malaria parasites which boost the increased demand for folates in pregnancy.
Dietary pattern of the pregnant women was one of the associ- where a statistically significant increase (p < .01) was found between the overall prevalence of anemia in multipara women (64.0%) compared to primigravida women (49.3%) (Mohammad & Salahat, 2012).
A previous study witnessed that the hemoglobin concentration decreases in grand multiparas compared with nulliparas (Gebremedhin, 2014

| Strength and Weakness of the study
Even if the current study had adequate sample and better response rate, there might be biases as it is institution-based study which may exclude most at-risk women. Hence, the occurrence of biases might have underestimated or overestimated the prevalence rates of anemia as well as the dietary intake of the women.

| CON CLUS IONS
The prevalence of anemia was mild public health concern in the current study based on the World Health Organization standard. The early and timely uptake of the cheaper, faster, better, and evidencebased focused antenatal care approach and iron folic acid supplementation was poor in the study area. Similarly, diet diversity which is considered to be the proxy indicator for micronutrient intake and consumption of iron-rich foods was found to be low among the pregnant women. Being from the lower wealth index and educational status categories were also main affecting factors to anemia.
Thus, the burden of anemia should be alleviated through improved uptake of the antenatal care services available in the health centers.
Economic and educational empowerments like livelihood promotion strategies have affirmative inputs. Diet diversity and intake iron-rich foods must be enhanced especially during pregnancy with effective nutrition education. To identify additional contributing factors for anemia, we recommend community-based study using additional variables like weighed food scores so as to precisely determine the dietary intakes and its impacts on anemia.

ACK N OWLED G EM ENTS
The current study did not receive any funds. The authors acknowledge the data collectors and respondents participated in the present study. They are also grateful for the School of Nutrition, Food Science and Technology, Hawassa University for supports given during the data collection processes.

CO N FLI C T O F I NTE R E S T S
All authors declare that they have no competing interests. The lead author affirms that this manuscript is accurate and transparent.

E TH I C A L A PPROVA L
The study conforms to the Declaration of Helsinki guidelines for human subjects and approval was given by Hawassa University Institutional Review Board (IRB). Prior to data collection, the district administrators were contacted with a description of the study and purposes and consent was taken. The nature of the study was fully explained to the respondent and oral consent was taken ahead of interview. The respondents who were anemic and who had not started treatment were provided iron tablets and counseled on recommended nutritional options.