Maternal dietary pattern and its association with birthweight in Northern Ethiopia: A hospital‐based cross‐sectional study

Abstract Birthweight is a useful public health measure of maternal health, nutrition, healthcare delivery, and child morbidity and mortality. Previous research did not focus on dietary patterns but rather on a single or a few foods or nutrients. This study aimed to assess the maternal dietary pattern and its association with birthweight in northern Ethiopia. A hospital‐based cross‐sectional study was conducted among 373 pregnant mothers in their third trimester of pregnancy who came to attend their routine antenatal care service. The food frequency questionnaire was collected from the previous week, and the birthweight data were collected from the medical records after delivery. Three maternal dietary patterns were identified; dietary pattern includes eggs, milk, milk products, and certain fruits and roots. Dietary pattern 2 includes certain vegetables, green leafy vegetables, vitamin A‐rich vegetables, pulses such as beans, peas, and chickpeas, and drinks like coffee, tea, and soda. Dietary pattern 3 includes meat, nuts, and grains such as teff, corn, wheat, and white flour. Dietary pattern 1 (β = 52.45, p = .03) and dietary pattern 2 (β = 66.76, p = .01), residency (β = 287.08, p < .001), a mid‐upper‐arm circumference of 21–23 cm (β = 187.10, p = .02), a mid‐upper‐arm circumference of >23 cm (β = 272, p = .01), and gestational age at delivery (β = 12.58, p = .004) were the factors significantly associated with increased birthweight. The maternal dietary pattern has a significant association with birthweight. The focus should be given to maternal dietary patterns to prevent suboptimal and high birthweight.

A single or a small number of foods or nutrients were the major focus of earlier investigations to determine the connection between maternal nutrition and birthweight (Slattery, 2010). The singlenutrient method, however, has several drawbacks (Northstone et al., 2008). Instead of eating isolated nutrients, people consume meals made up of a variety of foods; single-nutrient analysis does not take into account complex interactions between nutrients, and nutrient impacts are challenging to examine separately by singlenutrient analysis (Slattery, 2010).
Using dietary patterns rather than isolated nutrients would be better for understanding the gap in maternal and child morbidity, and mortality related to malnutrition (Chia et al., 2019). As the previous studies were from Western countries, there is limited data available regarding the association between maternal dietary patterns and neonatal birthweight in Africa, including in Ethiopia (Abubakari & Jahn, 2016). Evaluation and a better understanding of maternal dietary patterns will improve the birthweight outcomes, maternal and child morbidity, mortality, and complications, and improves the quality of care given in nutritional counseling for pregnant mothers.
This study was intended to assess the maternal dietary pattern and its association with birthweight in Northern Ethiopia.

| Key messages
• Complex interactions between nutrients are not taken into account by the single-nutrient analysis, and it is challenging to analyze nutrient impacts separately.
• According to our research, the food patterns of pregnant women in their third trimester of pregnancy can be best explained by three unique dietary patterns.
• A higher birthweight result was observed in women who ingested more eggs, milk and milk products, pulses, vegetables and roots, fruits, beverages, and sweets.
• For the majority of the time, food habits usually remain steady during pregnancy, and there is a strong correlation between maternal dietary pattern and birthweight.

| Study area and period
This study was conducted in Adigrat general hospital, a governmentowned general hospital and teaching center for Adigrat University.
It also serves as a referral hospital for surrounding catchment areas visit, 1188 on their fourth ANC visit, and a total of 2852 deliveries.
The study was conducted from March 1 to May 29, 2020.

| Study design and participants
A hospital-based cross-sectional study was conducted among pregnant women who were in their third trimester of pregnancy (>37 weeks of gestation) coming to attend their routine antenatal care service and their respective live-born baby medical records in Adigrat general hospital, Tigray, Ethiopia. Mothers with ultrasoundconfirmed multiple pregnancies, who were mentally unstable and critically ill, mothers with pregnancy-induced hypertension, and gestational diabetes mellitus were excluded from the study since they need special dietary intake.

| Sample size and sampling technique
The sample size was calculated using a single population proportion formula with the assumption of 95% CI and a 5% margin of error.
The proportion of pregnant mothers who ate an unhealthy dietary pattern, p = 28.3%, was taken (Abubakari & Jahn, 2016). With these presumptions, the total sample size become 312, and when a 15% non-response rate was added, the ultimate sample size increased to 373. A consecutive sampling technique was used to select eligible mothers coming for their routine antenatal care visit to the hospital.  (Persson et al., 1998).

| Data collection tool and procedure
Frequencies of consumption were taken by a recall of dietary intakes of the previous week from the data collection day. Anthropometric measurements of the mothers were assessed by using the midupper-arm circumference (MUAC) and weight.
A digital weight scale was used to measure maternal weight.
First, the scale was zeroed before stepping on them, the mother was asked to step onto the center of the scale and to stand still, then we waited until the weight was displayed and remained fixed in the display panel, and read aloud the weight to the nearest 0.1 kg. Weight before pregnancy was self-reported but if she does not remember, it was taken retrospectively from her medical record. MUAC was measured using a measuring tape and recorded accurately to the nearest 0.1 mm.
Laboratory measurement for hemoglobin level was also recorded from the laboratory result papers. Secondary data on the birthweight of their respective live births were taken from their medical record after delivery. Six data collectors who work at the health institution (midwifery, nurses, and health officers) and one supervisor collected the data. The birthweight of the newborn was the main outcome variable, the independent variables were the sociodemographic characteristics, reproductive characteristics, medical and risky behavior characters, and nutritional factors.

| Data management and analysis
The collected data were coded and checked for consistency and completeness up to the end of each data collection period.

| Reproductive health characteristics, medical, and risky behavioral factors
The mean birthweight of the newborn was 3125 g with a standard deviation of (±460) g.

| Maternal nutritional status
During this study period, 65 (18.36%) of the pregnant mothers were eating fast foods. One hundred eighty-eight (55.46%) of the pregnant mothers reported having received nutritional counseling from a health professional during the current pregnancy.
Regarding the feeding frequency, 285 (80.51%) of the mothers ate greater than three times a day, and concerning the anemia status of the mothers, 341 (96.33%) had a hemoglobin level of ≥11 g/dL.

| Dietary patterns
After removing foods that were consumed less than 25% of the time, the food items were grouped into 10 food groups. According to our research, the food patterns of pregnant women in their third trimester of pregnancy can be best explained by three unique dietary patterns. They are labeled as pattern 1, pattern 2, and pattern 3, and these patterns explained 44.9% of the variability in the dietary pattern of pregnant women. Food groups with high factor loadings (0.3) on dietary pattern 1 included eggs, milk, and milk products, certain fruits (banana, orange, avocado, papaya, guava, and mango), and roots (potato and carrot).
Dietary pattern 2 was characterized by high factor loadings for certain vegetables (green leafy vegetables and vitamin A-rich vegetables), pulses (beans, peas, and chickpeas), and drinks like coffee, tea, and soda. Dietary pattern 3 had higher loading factors for nuts, grains (teff, corn, wheat, and white flour), and meat. Dietary pattern 3 also had very low (negative) factor loadings for eggs, milk, and milk products, and pulses ( Table 4).
TA B L E 1 Sociodemographic characteristics of pregnant women receiving routine antenatal care service in Northern Ethiopia, 2020 (n = 354).

Maternal profile Categories Frequency (%)
Age, median (interquartile range), years 26 ( TA B L E 2 Reproductive health characteristics, medical, and risky behavioral factors of pregnant women receiving routine antenatal care service in Northern Ethiopia, 2020 (n = 354).

| Mean birthweight difference
There is a statistically significant difference in the newborn birth-

| Independent effect of dietary patterns on birthweight
Among the three extracted dietary patterns of the pregnant mothers during their third trimester of pregnancy, dietary pattern 1 and dietary pattern 2 were significantly associated with higher birthweight while dietary pattern 3 was not significantly associated.

| DISCUSS ION
This study tried to assess the maternal dietary pattern and its Although, until recently, there was little research on women's food habits during pregnancy, particularly in Sub-Saharan Africa, those studies did find several dietary patterns that were almost identical to those seen in our investigations (Tshotetsi et al., 2019). For instance, two dietary patterns that can be classified as health-conscious and non-health-conscious were found in a study of pregnant women in Ghana (Abubakari & Jahn, 2016).
Three dietary patterns-animal-based, plant-based, and grainbased-were discovered in another study conducted in Malawi among pregnant HIV-positive women (Ramlal et al., 2015). Four dietary patterns-animal-based, staple-based, recommended diet, egg and breakfast cereals, legumes, and vegetables-were also dis-

covered in a second study on the HIV-infected population in South
Africa (Annan et al., 2015). A study on Danish women also identified three categories of food habits: western, intermediate, and healthconscious patterns (Knudsen et al., 2008).

TA B L E 2 (Continued)
Another study among expectant New Zealanders found three patterns: traditional, fusion, and junk diets (Thompson et al., 2010).
But Mexican-American women (Wolff & Wolff, 1995) and Finnish women (Arkkola et al., 2008) and each identified seven eating patterns. In contrast, these studies employed a lower factor loading of 0.2 as opposed to 0.3, which is used by others, including the current study. This study is consistent with the majority of studies on maternal food habits that showed a connection between the patterns found and birthweight. According to this study, there is a substantial link between the two dietary patterns that include fruits, eggs, milk,

vegetables, and other protein-rich foods and an infant's birthweight
increase. This was corroborated by a study on Mexican-American women, which showed that nutrient-dense, protein-rich meals were linked to higher infant birthweight (Wolff & Wolff, 1995).
In addition, there were trends toward associations between a higher VFR (vegetable, fruit, and rice) pattern score and higher birthweight in a study conducted in China (Chia et al., 2019) similar to ours. Overall, it found that healthier diets classified as ones containing eggs, milk, fruits, and vegetables consistently produced superior results in terms of birthweight, as seen in this study (Chia et al., 2019). There are several potential explanations for the link between the increasing consumption of fruits, vegetables, and foods high in protein and birthweight.
Several essential minerals, including potassium, magnesium, dietary fiber, folate, and vitamins A and C, are mostly found in fruits and vegetables (Knudsen et al., 2008). A diet high in fruits and vegetables supports healthy immunological and placental processes, both of which are crucial for the development of the fetus (Kliegman et al., 2016). Alternatively, increased fruit and vegetable intake may simply be a marker for healthier dietary patterns or a healthier lifestyle that could not be fully adjusted in this study. While other studies have shown an association between cereal and red meatbased diets with lower birthweight outcomes in infants (Knudsen et al., 2008), our study has identified no significant association. This could be due to a small variation between the mothers in the intake of cereals because almost all mothers in our study consumed cereals daily. Although in the studies cereal was consumed on a daily base by almost all mothers similar to our study, the foods were adjusted using portion size, unlike our study. However, our dietary TA B L E 3 Nutritional status of pregnant women receiving routine antenatal care service in Northern Ethiopia, 2020 (n = 354

TA B L E 5
One-way analysis of variance for birthweight by different maternal profiles among pregnant women receiving routine antenatal care service in Northern Ethiopia, 2020 (n = 354 Note: The mean difference is significant at the *p < .05 and **p < .001. pattern 1 had a very low (negative) loading factor for grains and sweetened beverages and alcohol, which is associated with higher birth and weight outcomes.
High adherence to dietary patterns 1 and 2 was found to significantly increase the likelihood of having an infant with a higher birthweight. A 1 unit increase in the loadings of dietary pattern 1 was linked TA B L E 7 The association between maternal factors and birthweight among pregnant women receiving routine antenatal care service in Northern Ethiopia, 2020 (n = 354 to an increase in birthweight of 52.45 g, while a 1 unit increase in the loadings of dietary pattern 2 was linked to an increase in birthweight of 66.76 g. Similar results were found in earlier investigations, which indicated that residence also appeared to be significantly related to birthweight (Mengesha et al., 2017). In a study conducted in Tigray, mothers living in a rural area were at increased risk of delivering lowbirthweight infants (Mengesha et al., 2017)  These inadequacies could put these women at risk of receiving subpar prenatal care and related services. Additionally, these ladies may not be getting enough nourishment because they live in a remote area, as most studies have found that rural residents are not getting enough nutrition (Islam, 2015). Additionally, there was a strong correlation between birthweight and gestational weight increase. This was supported by a study done in Mekelle private clinics, which has shown that a 1 unit increase in weight gain during pregnancy was associated with a 0.48 increase in birthweight (Tela et al., 2019).
This could be due to the major contribution weight gain during pregnancy makes to intrauterine fetal growth, which is an important measure of prenatal nutrition (Kliegman et al., 2016). Similar to this, a within-family analysis conducted in the United States found that inadequate GWG increases the likelihood of low birthweight, whereas excessive weight gain considerably raises the probability of having high-birthweight babies (Goldstein et al., 2017). This similarity may result from the fact that no single item can satisfy all of a person's nutritional needs, therefore the more food categories one consumes daily, the more likely they are to meet those needs (Dewey, 2016).
Therefore, a sufficiently diverse diet may reflect nutrient adequacy (Abubakari & Jahn, 2016) and increased birthweight. In addition, gestational age at delivery was also significantly associated with birthweight. Similar to our study, prematurity was significantly associated with birthweight in previous findings from Iran (Chaman et al., 2013) and Kenya (Muchemi et al., 2015). According to a study conducted in the Tigray region of Ethiopia, preterm delivery was nearly three times as significantly associated with macrosomia as post-term birth was with low birthweight (Mengesha et al., 2017).
Additionally, in a study done in Northern Ghana, longer gestation was linked to heavier babies; each additional week of gestation was linked to a 0.141 standard-unit increase in birthweight (Abubakari & Jahn, 2016).
The strength of this study is that it assessed the effect of food interactions on birthweight which a single nutrient analysis may not be able to identify. In addition, we paid attention to environmental, behavioral, and lifestyle factors, as well as other nutritional and pregnancy-related factors, in addition to sociodemographic characteristics. It is also important to note the study's eligibility requirements, which call for healthy women who do not have any of the conditions listed in the protocol (such as diabetes mellitus and hypertension) and do not require particular dietary regimens.
Dietary intake was evaluated before the outcomes were known, which is another significant strength.
The limitation of this study is predominantly that the FFQ used to collect data was not an FFQ specifically prepared for this study population. Each population has its dietary habits. Age, socioeconomic standing, racial or ethnic origin, cultural practices, and access to food can all affect them (Northstone et al., 2008).
Therefore, for this limitation, a modified Helen Keller FFQ for vitamin A assessment was used. Additionally, there are flaws in every nutritional assessment technique. FFQs are inaccurate and prone to underreporting.
In addition, because we only looked at the previous week's food intake, we cannot guarantee that the data are indicative of dietary patterns throughout the entire pregnancy. But prior research has indicated that pregnant women's overall food habits stay the same (Northstone et al., 2008). Thirdly, as information on portion size was not gathered in the current study and dietary consumption was mainly based on the frequency of food intake, we were unable to account for total calorie intake. Last but not least, despite making multiple statistical adjustments, residual confounding associated with fetal growth and diet during pregnancy cannot be completely ruled out.

| CON CLUS IONS
In conclusion, three dietary patterns were identified; of those patterns, dietary pattern 1 and dietary pattern 2 were significantly associated with birthweight. Women who consumed more eggs, milk and milk products, pulses, vegetables and roots, fruits, beverages, and sweets had an increased birthweight outcome. Therefore, intrauterine growth conditions in the third trimester might be improved by adopting certain dietary patterns during pregnancy, especially dietary patterns dense with egg, milk, and milk products, fruits, vegetables and roots, and pulses.

ACK N OWLED G M ENTS
The authors thank all pregnant women who took the time to complete the survey.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets generated and analysed during the current study are available from the corresponding author on a reasonable request.