Women in the reproductive age group and children are most vulnerable to malnutrition due to low dietary intakes, inequitable distribution of food within the household, improper food storage and preparation, dietary taboos, infectious diseases, and care [1] . Under nutrition threatens both the health and survival of mothers and their infants, because it increases their susceptibility to life-threatening diseases and their risk of dying, especially during childbirth [2] . Gaining too much weight and obesity is also associated with an increased risk of medical and puerperal complications, including hypertension, preeclampsia, gestational diabetes mellitus, thrombophlebitis, labor abnormalities including prolonged second-stage labor, delivery after 42 weeks of gestation, and cesarean delivery. Operative complications among obese women undergoing cesarean delivery include increased blood loss, prolonged operative time, and increased rates of postoperative infection [3] . Excessive weight gain may also cause problems for the baby. Large babies make vaginal deliveries difficult, increasing the risk for cesarean section. Overweight babies may have an increased risk for health problems later in life (e.g., obesity, diabetes, and cardiovascular problems) [4] .

Worldwide, estimated 852 million people are undernourished with most (815 million) living in developing countries [5] [6] . Poverty is the main underlying cause of malnutrition and its determinants [7] . The degree and distribution of Protein Energy Malnutrition (PEM) in a given population depend on many factors―the socioeconomic situation, level of education and sanitation, the season and climate conditions, food production, cultural and religious food customs, breastfeeding habits, prevalence of infectious diseases, the existence and effectiveness of nutrition programs and the availability and quality of health services [6] [8] .

Under-nutrition or low Body Mass Index (BMI) continues to be a major health burden in addition to the emergence of new competing public health priority (high BMI) in developing countries [9] . It is globally the most important risk factor for illness and death with hundreds of millions of pregnant women and young children particularly affected [10] . Poor nutrition in pregnancy in combination with infections is a common cause of maternal and infant mortality and morbidity, low birth weight and Intrauterine Growth Retardation (IUGR) [11] .

Ethiopia is the second most populous country in Africa, and is a home to about 74 million people, most of who live in rural areas [12] . According to the study by the Ethiopian Ministry of Economic Development and Cooperation, 50 percent of the Ethiopian population are living below the food poverty line and cannot meet their daily minimum nutritional requirement of 2200 calories [13] . The country’s maternal health statistics are grim: the number of maternal deaths is fifth highest in the world [14] and the maternal mortality ratio, estimated to be 673 per 100,000 live births, is extremely high [15] . The extent to which malnutrition contributes to these maternal deaths is unknown. However, it remains one of the world’s highest priority health issues not only because its effects are so widespread and long lasting but also because it can be eradicated [14] .

Researches have been done in the countries where pre-pregnant BMI determination is difficult; 1^st visit pregnancy BMI has been used to determine the nutritional status of the mothers and assess pregnancy BMI risk factors (like age, parity, economic status). In most developing countries including Ethiopia women do not tell their baseline weight and height; in this situation the need to determine weight and height and thereby pregnancy BMI at 1^st Antenatal Care (ANC) booking becomes evident so as to guide the counseling need on nutrition during subsequent ANC visits [10] [16] - [21] . Additionally, adverse outcomes of abnormal pregnancy BMI such as pre-eclampsia, induced labor, postpartum hemorrhage, preterm delivery, birth weights less than 2500 g and birth weights >4000 g were documented in other researches [22] .

However, such study was not carried out in Ethiopia. The authors believe that, in order to reduce (prevent) the adverse outcomes of abnormal pregnancy BMI, the need to identify groups of pregnant women at greater risk of developing abnormal BMI and identify its modifiable risk factors would have paramount importance. Therefore, this study was proposed to determine and compare prevalence and identify important independent socio-demo- graphic/economic risk factors of low and high BMI among urban versus rural pregnant women at their first visit or booking for ANC in governmental health centers of Jimma Zone, Southwest Ethiopia.

A total of 1546 pregnant mothers (770 urban and 776 rural residents) who fulfilled the inclusion criteria were included in the study. The response rate varies for different variables of the study ranging between 93.9% for household expenditure and 100% for most of the variables like gestation period, autonomy, health problem at booking, season at booking, and for the composite variable BMI (dependent variable of the study).

This study determined the prevalence and identify important independent socio-demographic/economic risk factors of low and high BMI among urban versus rural pregnant women at their first visit or booking for antenatal care in governmental health centers of Jimma Zone, Southwest Ethiopia.

Similar to other developing countries [9] , the pregnant mothers visited health institutions and booked for ANC in the late weeks of gestation. Majority (59.7%) of the pregnant mothers, urban (60.8%) and rural (58.6%) residents, came for their 1^st ANC visit in the second trimester of pregnancy. However, only 9.4% of the pregnant mothers, 8.7% of urban and 10.2% of rural residents, came in their 1^st trimester for the 1^st ANC booking.

Majority of, 60.8% urban and 60.2% rural resident, pregnant mothers had normal BMI at the time of their booking for ANC (on static/outreach basis) in the selected governmental health centers (Woredas) of Jimma Zone. Prevalence of low BMI was 24.8% among all of the studied pregnant mothers, lesser (22.3%) in urban than rural (27.2%) residents; whereas the prevalence of high BMI was 14.7% among all of the studied pregnant mothers, higher in urban (16.9%) than rural (12.6%) residents, at the time of booking. Compared to these findings, lower prevalence of PEM was reported in the study conducted on pregnant women recruited from antenatal clinics in Khon Kaen province (15.1% in 1^st trimester) in Thailand in 2003 [19] , and Dar es Salaam, urban Tanzania, (overall 2.9% and 5.6% in 1^st trimester) in 2003 and 2004 [9] . In the Nigeria’s study [18] , the mean BMI was significantly lower in rural than urban pregnant mothers. On the other hand, the prevalence of BMI > 25 (overweight and obesity in <14 weeks of gestation) in Dar es Salam (30.9% in 2004) was greater than the findings of the current study of our settings (12.3% in the urban and 11.6% in the rural mothers in ≤24 weeks of gestation). These differences might be due to socio-economic, cultural and genetic differences. Also, statistically

Normal BMI is the reference for BMI categories; COR = Crude Odds Ratio; AOR = Adjusted Odds Ratio; Wald test: ^*p < 0.05; ^**p < 0.01;^*** p < 0.001; The standard error for the “b” coefficients of all independent variables of the multivariate model was not larger than 2.0; The categorical variables of age, education, parity, gestation week, autonomy, household expenditure, solid waste disposal system, safety of drinking water, latrine facility, health problems at booking and season at booking were adjusted in the final model; The cases to variables ratio (710:19) was 37.4 to 1; Model Fitting Information: the probability of the final model chi-square (94.89) was 0.000; The likelihood ratio test was significant only for Education (x² = 12.31, p < 0.05), gestation week (x² = 35.46, p < 0.001) and household expenditure (x² = 14.84, p < 0.05); The proportional by chance accuracy rate was 0.218² + 0.618² + 0.163² = 0.456; The proportional by chance accuracy criteria was 57.4%; The overall correct classification accuracy rate was 63.4% (low BMI group 3.2%, normal BMI group 97.9% and high BMI group 12.9%), which is not greater than 2%, when compared to the accuracy rate after omitting outliers (63.8%). Hence, the model with all cases was interpreted in this analysis.

Normal BMI is the reference for BMI categories; COR = Crude Odds Ratio; AOR = Adjusted Odds Ratio; Wald test: ^*p < 0.05; ^**p < 0.01;^*** p < 0.001; The standard error for the “b” coefficients of all independent variables of the multivariate model was not larger than 2.0; The categorical variables of age, education, parity, gestation week, autonomy, household expenditure, solid waste disposal system, safety of drinking water, latrine facility, health problems at booking and season at booking were adjusted in the final model; The cases to variables ratio (683:19) was 35.9 to 1; Model Fitting Information: the probability of the final model chi-square (69.36) was 0.001; The likelihood ratio test was significant only for gestation week (x² = 19.58, p < 0.01) and household expenditure (x² = 18.70, p < 0.01); The proportional by chance accuracy rate was 0.274² + 0.615² + 0.111² = 0.466; The proportional by chance accuracy criteria was 58.3%; The overall correct classification accuracy rate was 64.1% (low BMI group23.5%, normal BMI group 93.8% and high BMI group 0%), which is not greater than 2%, when compared to the accuracy rate after omitting outliers (64.3%). Hence, the model with all cases was interpreted in this analysis.

significant difference was observed in the distribution of the proportions of BMI categories of the respondent mothers by their place of residence (p < 0.05) and different Woredas (health centers) (p < 0.001) included in the current study.

Compared to the PEM figure of the Thailand’s and Dar es Salaam’s studies [9] [19] , the proportions of low BMI of the pregnant mothers at booking for ANC (on static/outreach basis) in the selected health centers or Woredas (Kersa (37.6%), Omonada (33.3%), Mana (28.5%), Gera (19.2%), Shebe Sombo (17.4%)) were high except for one Woreda (Seka Chekorsa (12.5%)). Along this, high BMI was found in considerable proportions of the pregnant mothers residing in these Woredas (Gera (23.5%), Omonada (22.1%), Shebe Sombo (14.7%), Seka Chekorsa (10.2%), Mana (9.4%) and Kersa (8.5%)). This has a great implication for the need of designing and implementing abnormal BMI prevention and controlling strategies in the pregnant women of the studied areas/ Jimma zone at large.

In our study, multiple multinomial logistic regression model was employed separately for the urban and rural mothers to identify independent predictors of their BMI status (categories). In both models categorical variables of age, education, parity, gestation week, autonomy, household expenditure, solid waste disposal system, safety of drinking water, latrine facility, health problems at booking and season at booking were adjusted.

In the urban pregnant mothers, only primary and above education level and 3^rd trimester gestation period of mothers were found to be important independently associated factors for high BMI of the urban pregnant mothers at the time of their booking for ANC. The urban pregnant mothers with primary and above education level were more likely to have high BMI than the illiterates. Similarly, the mothers in higher gestational weeks (3^rd trimester) were more likely to have high BMI than in lower gestational weeks (2^nd trimester). This finding is consistent with the finding of a study done in Nigeria that showed significantly lower mean BMI and higher PEM prevalence among the less educated (no formal and primary education) pregnant women [18] , since the category in our study included the mothers with above primary level of education too. It is also supported by the finding of a cohort study done in Rio de Janeiro state, Brazil, showed that, in the first trimester, pregnant women who had a higher education level were more likely to gain more weight than recommended [25] . In Recife, Pernambuko, Brazil, a study also reported the association of excessive weight gain with schooling of pregnant women in the third trimester [26] . In the later study, excessive weight gain was common in all categories of baseline nutritional status, but was more frequent in the second trimester among women who were already overweight or obese upon entering pregnancy [26] . Inconsistent to these findings, in the Dar es Salam’s study, obesity had no significant association with an increase in gestation weeks [9] .

In the rural pregnant mothers, only higher monthly household expenditure and 3^rd trimester gestation period were found to be important independently associated important factors for low BMI of the rural pregnant mothers at the time of their booking for ANC after controlling for the confounding effect of the other variables. Consistent to Dar es Salam’s study [9] , low BMI was less likely to happen in third trimester of pregnancy than second trimester. But, low BMI was more likely to occur in mothers who had higher monthly household expenditure than those spending less than or equal to 350 Eth.birr per month. In contrary to this, in the Dar es Salam’s study food purchasing power (money spent on food) was positively associated with obesity and inversely related to underweight in pregnant women [9] . The household expenditure might have been not directly related to purchase of food and dietary consumption of the mothers or households of the rural settings of our study.

Furthermore, unlike to our study, in the Dar es Salam’s study [9] , obesity was positively associated with age, parity, and inversely with health problem (HIV infection). underweight was positively associated with HIV status. This indicates the need for more strong studies to identify factors affecting BMI and weight gain during pregnancy and take evidence based measures to ensure the wellbeing of mothers and children of our settings in the future.

Generally, the study used adequate sample size and produced generalizable findings to the target population. However, the weaknesses of using secondary data (extracted weight and height from ANC cards of mothers) was observed in the study. The ANC records had some incomplete information which resulted a considerable number of missing values, for instance for LMP, and, hence, exclusion of such items of the questionnaires from the analysis were made. The BMI at booking (though, not specific measure of nutritional status of pregnant women) was used in the study regardless of the extent of gestation week retrieved also from the routine ANC cards of the mothers by adjusting it in the multivariate multinomial logistic regression models, which is an acceptable procedure in such cases.

In conclusion, in this study, very low proportion of the urban and rural pregnant mothers booked for the first ANC in their 1^st trimester of pregnancy, while the majority booked in the 2^nd trimester. A considerable proportion of both urban and rural pregnant mothers had low BMI as well as high BMI. The prevalence of low BMI was higher in rural than urban pregnant mothers, whereas high BMI was higher in the urban than rural pregnant mothers. The decreasing order of the Woredas by proportions of low BMI of the pregnant mothers was Kersa, Omonada, Mana, Gera, Shebe Sombo, and Seka Chekorsa; but, the order by the proportions of high BMI of the respondents was Gera, Omonada, Shebe Sombo, Seka Chekorsa, Mana, and Kersa. Moreover, primary and above education level and 3^rd trimester gestation were found as important positively and independently associated factors for high BMI of the urban pregnant mothers; while, higher monthly household expenditure (positively) and 3^rd trimester gestation period (negatively) were found as important independently associated factors for low BMI of the rural pregnant mothers at the time of their booking for ANC. Hence, the health bureaus of the zone/Woredas/health facilities of the study area should strengthen delivery of health education to the community to enable pregnant women to book in health facilities for the first ANC in the recommended time (1^st trimester of pregnancy). The health workers in health facilities should be able to determine/document the BMI status of every pregnant woman at booking for ANC, monitor weight gain and offer relevant health and nutritional advise/care to clients in all of the subsequent visits during pregnancy. FMOH/regional/all concerned governmental and non-governmental organizations should design a strategy to introduce the culture of monitoring nutritional status (BMI status) in women of reproductive age in their pre-pregnancy and monitoring body weight in their pregnancy periods in the study area and the country at large. Finally, undertaking further studies using strong design on the issue is also advisable in different settings of the country at large.

We, authors, declare that we do not have competing interests.

C.H., T.D., and B.T. participated in the conception and development of proposal of the study, training of data collectors, pretest of instrument, coordination and supervision of data collection, data entry, cleaning, analysis and interpretation of the data. C.H. and B.T. drafted the manuscript. All authors involved in editing the first draft, and read and approved the submission of the final manuscript for publication.

We would like to thank Jimma University for funding the research project; authorities of Jimma Zone and study Woredas’ health offices/health facilities, the study participants (pregnant mothers) and the data collectors (health facility staffs) for their unreserved cooperation during the data collection process in their settings.

Chernet Hailu,Tariku Dejene,Bosena Tebeje, (2015) Prevalence and Socio-Demographic/Economic Risk Factors of Low and High Body Mass Index of Urban versus Rural Pregnant Women at Booking for Antenatal Care in Governmental Health Centers of Jimma Zone, Southwest Ethiopia. Open Access Library Journal,02,1-17. doi: 10.4236/oalib.1101672

BMI Low Body Mass Index

ANC Antenatal Care

COR Crude Odds Ratio

AOR Adjusted Odds Ratio

Eth.birr Ethiopian Birr

PEM Protein Energy Malnutrition

IUGR Intrauterine Growth Retardation

LMP Last Menstrual Period

SPSS Statistical package for Social Sciences

FDROE Federal Democratic Republic of Ethiopia

MOPED Ministry of Planning and Economic Development

WHO World Health Organization

CSA Central Statistics Authority