Higher Dietary Intake of Animal Protein Foods in Pregnancy Is Associated with Lower Risk of Adverse Birth Outcomes

ABSTRACT

Background

The prevalence of adverse birth outcomes is highest in resource-limited settings such as sub-Saharan Africa. Maternal consumption of diets with adequate nutrients during pregnancy may protect against these adverse outcomes.

Objectives

The objective was to determine the association between maternal dietary consumption of animal source foods (ASFs) and the risk of adverse birth outcomes among HIV-negative pregnant women in Tanzania.

Methods

Using dietary intake data from 7564 HIV-negative pregnant women, we used Poisson regression with the empirical variance (generalized estimating equation) to estimate the RR of adverse birth outcomes—preterm birth, very preterm birth, small for gestational age (SGA), low birth weight (LBW), stillbirth, and neonatal death—for higher and lower frequency of ASF intake.

Results

Median daily dietary intake of animal protein was 17 g (IQR: 1–48 g). Higher frequency of ASF protein intake was associated with lower risk of neonatal death (quartile 4 compared with quartile 1; RR: 0.59; 95% CI: 0.38, 0.90; P-trend = 0.01). Higher fish intake was associated with lower risk of very preterm birth (high tertile compared with low; RR: 0.76; 95% CI: 0.58, 0.99; P-trend = 0.02). Any meat intake was protective of preterm birth (RR: 0.73; 95% CI: 0.65, 0.82; P < 0.001), very preterm birth (P < 0.001), LBW (P < 0.001), and neonatal death (P = 0.01) but was associated with increased risk of SGA (RR:1.19; 95% CI: 1.01, 1.36; P = 0.04). Any egg intake was protective of very preterm birth (RR: 0.50; 95% CI: 0.31, 0.83; P = 0.01) as compared with no egg intake. Finally, any dairy intake was associated with lower risk of preterm birth (RR: 0.82; 95% CI: 0.68, 0.98; P = 0.03) and very preterm birth (RR: 0.53; 95% CI: 0.34, 0.84; P = 0.01).

Conclusions

Higher frequency of dietary intake of ASF is associated with lower risk of adverse birth outcomes in urban Tanzania. Promoting prenatal dietary intake of ASF may improve birth outcomes in this region and similar resource-limited settings

Introduction

Optimal fetal development has been linked to multiple lifelong benefits, such as improved school performance, improved health during adolescence (especially among girls), improved health in adult life, increased productivity and economic gains, and reduced burden of infectious diseases (1). Prevalence of suboptimal fetal development is highest in resource-limited settings, with the highest prevalence of prematurity, stillbirth, and small for gestational age (SGA) reported in south Asia and sub-Saharan Africa (SSA) (2, 3).

In 2015, an estimated 20.5 million live births were low birth weight (LBW), 91% of which were from low- and middle-income countries (LMICs), mainly southern Asia (48%) and SSA (24%) (4). Worldwide, an estimated 32.4 million infants are born SGA (5), and SGA fetuses are up to 4 times more likely to be stillborn (6). A recent pooled analysis of Tanzanian infants estimated SGA to be at 17%, preterm at 15%, and LBW at 7% (7); preterm and SGA newborns in this cohort had an increased risk of mortality during infancy (7).

Maternal undernutrition is highly prevalent in resource-limited countries in SSA and south Asia and is recognized as a key determinant of poor birth outcomes (8). Regulation of maternal dietary protein intake during pregnancy is essential for proper embryonic survival, growth, and development (9). Specific amino acids are required for certain processes involved in pregnancy, including implantation, placental growth and angiogenesis, and the transfer of nutrients from mother to fetus (9).

Animal-based food products in general contain the highest amount of protein per unit energy, and protein derived from animal foods is considered the best-quality protein (10). Adequate dietary intake of animal source protein foods is critical in pregnancy due to their complete amino acid profile; contents and bioavailability of lysine, sulfur amino acids, and threonine; and associated insulin-like growth factors, iron, zinc, and vitamin B-12 (11–15). However, dietary intake of protein during pregnancy is often low in LMICs, where dietary protein sources are mainly limited to cereals and, to a much lesser extent, high-quality animal sources (10).

Several studies in LMICs have examined the determinants of adverse birth outcomes (11–17). Yet, there is a paucity of literature on the relationship between maternal dietary intake and birth outcomes in these settings. To fill this knowledge gap, we examined the relations of maternal animal protein dietary intake with preterm birth, LBW, SGA, stillbirth, and neonatal death, utilizing data from HIV-negative pregnant women who were enrolled in a multivitamin supplementation randomized controlled trial in Tanzania.

Methods

Study design and population

Between August 2001 and July 2004, pregnant women who attended prenatal clinics in Dar es Salaam were enrolled in a trial assessing the effects of multivitamin supplements on birth outcomes. Eligibility requirements included a negative test result for HIV infection, a plan to stay in the city until delivery and for 1 y thereafter, and gestational age between 12 and 27 wk according to the date of the last menstrual period. The women were randomly assigned to receive a daily oral dose of a multivitamin supplement or placebo from the time of enrollment until 6 wk after delivery (n = 8468). Multivitamin supplements included 20 mg thiamin, 20 mg riboflavin, 25 mg vitamin B-6, 100 mg niacin, 50 μg vitamin B-12, 500 mg vitamin C, 30 mg vitamin E, and 0.8 mg folic acid. All study participants were provided with standard prenatal care, including daily doses of elemental iron (60 mg), folic acid (0.25 mg), and sulfadoxine-pyrimethamine for intermittent preventive treatment of malaria (18). A detailed description of the parent trial design and follow-up procedures has been published elsewhere (18). This trial was registered at clinicaltrials.gov as NCT00197548.

Data collection and study variables

Study women completed a baseline questionnaire that included their obstetric history and sociodemographic characteristics. Blood counts were obtained at baseline and 6 weeks postpartum. Baseline maternal anthropometric measurements including height and weight were collected at enrollment (i.e., between 12 and 27 weeks of gestation). Women who did not come to the study clinic for their monthly appointments were visited at home when possible and asked to come to the clinic if their condition allowed.

Full-time research midwives attended to the women at delivery. Newborns were weighed immediately after birth, with the weights of the infants and the placentas measured to the nearest 10 g. For this analysis, maternal BMI was based on weight in kilograms and height in meters at enrollment (kg/m²). The Filmer-Pritchett wealth index was computed based on household ownership of assets (19).

Dietary assessment

Dietary intake was assessed by 24-h recalls during monthly prenatal clinic visits from 8 weeks after enrolment until 36 weeks of gestational age (20). Interviews were conducted by research nurses trained on food and nutrition assessment. The multiple-pass recall approach was used, wherein pregnant women were asked to remember and freely report all foods and beverages consumed in the preceding day. Respondents were guided to recall eating occasions and times for each food consumed. This was accompanied by probing and detailed questions to determine the portion size of food intake, aided by a locally adopted food atlas (20). The Tanzania Food Composition Tables (21) were used to determine nutrient content and quantity. In this study, 24-h recalls with energy consumption between 400 and 5000 kcal and not >200 g of protein intake were considered plausible.

Outcome definitions

The study outcomes were as follows: preterm birth (before 37 wk of gestational age), very preterm birth (before 34 wk of gestational age), LBW (<2500 g), SGA (birth weight <10th percentile for gestational age according to INTERGROWTH standards) (22), stillbirth (fetal death after 28 completed weeks of pregnancy), and neonatal death (within the first 28 d of life). We include women with singleton births.

Statistical analyses

Baseline data on maternal characteristics were used to characterize the cohort. Continuous variables were presented as mean and standard deviation and as median and interquartile range for skewed distribution. Categorical variables were summarized as percentages.

Mean daily dietary intake of animal protein (fish, meat, eggs, and dairy foods) was estimated as the average of multiple 24-h dietary recalls. For each person and each subset of foods, we counted how many servings she had for these foods in each 24-h recall. We then averaged these frequencies over the index pregnancy. Higher frequency of consumption was used as a proxy for total protein intake in the study. The mean daily frequency of total dietary intake of fish, meat (including chicken, beef, pork, goat), eggs, and dairy foods was the exposure variable for this analysis. The variable for the frequency of animal protein consumption encompassed the mean number of times that all animal protein foods were consumed daily, categorized into groups based on quartiles: ≤25th percentile (intake: ≤0.50 times/d), 25.1th–50th percentile (intake: >0.50 to 1.00 times/d), 50.1th–75th percentile (intake: >1.00 to 1.67 times/d), and > 75th percentile (intake: >1.67 times/d). Since large fractions of the women did not report eating some subtypes of foods, we had to make fewer categories for them. The fish variable was categorized into tertiles: low (0 times/d), medium (>0 to <0.67 times/d), and high (≥0.67 times/d). The meat, eggs, and dairy variables were dichotomized (any intake, no intake).

In a sensitivity analysis (Supplemental Table 1), we computed the daily total animal protein intake for women in grams per day. The raw data allow us to compute nutrient intake based on food and amount consumed, using the Tanzania Food Composition Tables. We classified the daily total animal protein intake into quartiles: quartile 1 (<8.8 g/d), quartile 2 (8.8 to <24.3 g/d), quartile 3 (24.3 to <43.2 g/d), and quartile 4 (≥43.2 g/d).

We collected data on protein intake in grams per day but used the daily frequency of animal protein intake as an indicator that is easily translated to nutrition recommendations and policy. In LMICs, it is often whether a pregnant woman consumed animal source foods that matters, and the grams of intake may not be the most important factor. Furthermore, for some sources of protein, such as meat, egg, and dairy, consumption was low in the study population, so deriving an indicator of any consumption was most important. In the sensitivity analysis using grams of protein intake, our findings were unchanged for most outcomes, which suggests that frequency of consumption is a reasonable metric.

We modeled the binomial outcomes using Poisson regression with the empirical variance (generalized estimating equation) (23) to estimate the RR of each outcome for the higher levels of daily frequency of total animal protein, fish, meat, eggs, and dairy intake as compared with the lowest level of intake. All covariates were adjusted for in multivariate regression analysis: maternal age, mother's education level, baseline hemoglobin level, parity, baseline BMI, food expenditure per day, energy intake in kilocalories per day, and a Filmer-Pritchett wealth index. Additionally, in multivariate regression, we included a variable derived from the frequency of daily intake of other main food groups: pulses, seeds and nuts, vegetables, fruits, cereals, roots tubers and bananas, and oils and fats. In sensitivity analysis, we evaluated associations between daily total animal protein intake for women in grams per day (quartiles) and poor birth outcomes. For variables missing >1% of the observations, missing data were retained in the analysis using the missing indicator method (24). Variables with missing data were as follows: mother's education (n = 1 of 7564), parity (0.1%), Filmer-Pritchett wealth index (0.1%), baseline BMI (13%), and baseline hemoglobin concentration (14%).

We estimated risk ratios and their corresponding 95% CIs. P < 0.05 was considered for statistical significance. We evaluated the associations of the frequency of 1) animal protein food intake across quartiles and 2) fish intake across tertiles with adverse birth outcomes by testing for trends across order groups .

In sensitivity analysis, we evaluated associations between daily total animal protein intake for women in grams per day (quartiles) and poor birth outcomes. All analyses were performed using the Statistical Analysis Systems statistical software package version 9.4 (SAS Institute Inc.).

Ethical approval

This study was conducted according to the guidelines in the Declaration of Helsinki, and all procedures involving human subjects/patients were approved by the institutional review boards of the Harvard TH Chan School of Public Health Human Subjects Committee, the Muhimbili University of Health and Allied Sciences Committee of Research and Publications, and the Tanzanian National Institute of Medical Research. Written informed consent was obtained from all patients.

Results

Of the 8468 pregnant women enrolled in the trial, 7564 (89%) had birth outcome data and plausible 24-h diet recall data and were included in this analysis. Mean ± SD maternal age was 25.2 ± 5.1 y (Table 1) and gestational age at recruitment was 21.2 ± 3.5 wk. Most women (67%) had received 5–7 y of formal education, and 33% had a baseline hemoglobin concentration ≥11 g/dL. Most women (61%) reported spending >500 Tanzanian shillings on food per day (1250 Tanzanian shillings were ∼1 US dollar at the time of the study).

TABLE 1.

Characteristics of the study participants: HIV-negative pregnant women in Dar es Salaam, Tanzania (N = 7564)

	Women
Characteristic	n	% or median [IQR]	Missing, n (%)
Multivitamin regimen			0 (0)
No (placebo)	3775	49.9
Yes	3789	50.1
Age, 1 y		25.2 ± 5.1	0 (0)
Mother's education, y			1 (0)
0–4	847	11.2
5–7	5054	66.8
8–11	1279	16.9
≥12	383	5.1
Parity			7 (0.1)
0	3403	45.0
1	2098	27.8
2	1123	14.9
≥3	933	12.4
Baseline BMI, kg/m2			954 (12.6)
<22	1773	26.8
22–24.9	2290	34.6
25–29.9	1916	29.0
≥30	631	9.6
Baseline hemoglobin, g/dL			1045 (13.8)
<8.5	767	11.8
8.5–10.9	3586	55.0
≥11	2166	33.2
Food expenditure,2 Tshs/d			587 (7.8)
>500	4213	60.4
≤500	2764	39.6
Filmer-Pritchett wealth index			7 (0.1)
< Median	3585	47.4
≥ Median	3972	52.6
Energy, kcal/d	7564	2225 [1830–2640]	0 (0)
Protein, g/d	7564		0 (0)
Total		59 [41–88]
Animal		17 [1–48]
Meat		0 [0–4]
Fish		0 [0–17]
Egg		0 [0–0]
Dairy		0 [0–0]
Vegetable		37 [28–48]

¹Mean ± SD.

²1250 Tanzanian shillings were ∼1 US dollar at the time of the study.

The mean number of 24-h diet recalls per woman was 2.7 ± 1.1. The median energy intake per day was 2225 kcal/d (IQR: 1830–2640); the median total protein intake per day was 59 g/d (IQR: 41–88); and the median animal protein intake per day was 17 g/d (IQR: 1–48).

The mean gestational age at delivery was 39 ± 3 wk, with 1093 (14.9%) preterm births (<37 wk) and 296 (4.0%) very preterm births (<34 wk). The mean birth weight was 3142 ± 496 g, with 447 (6.3%) births being LBW (<2500 g). There were 601 (8.7%) SGA infants, 235 (3.1%) stillbirths, and 174 (2.5%) neonatal deaths.

When compared with women in the lowest quartile, women in higher quartiles of the animal protein intake category trended toward lower risk of very preterm births in adjusted models (quartile 4 compared with quartile 1; RR: 0.80; 95% CI: 0.57, 1.12; P-trend = 0.07) (Table 2). Women with higher animal protein intake were less likely to have infants who died in the neonatal period (RR: 0.59; 95% CI: 0.38, 0.90; P-trend = 0.01). Dietary animal protein intake was not significantly associated with preterm birth, LBW, or stillbirth.

TABLE 2.

Univariable and multivariable associations of daily frequency of animal protein intake with adverse pregnancy outcomes¹

	Daily frequency of total animal protein intake, RR (95% CI) 2
Outcome	Q1	Q2	Q3	Q4	P-trend
Preterm delivery
n	364/2231	321/2156	131/1131	277/1800
Crude	Ref	0.91 (0.79, 1.05)	0.71 (0.59, 0.86)	0.94 (0.82, 1.09)	0.13
Adjusted	Ref	0.93 (0.81, 1.07)	0.75 (0.61, 0.90)	1.00 (0.85, 1.18)	0.52
Very preterm delivery
n	113/2231	90/2156	29/1131	64/1800
Crude	Ref	0.82 (0.63, 1.08)	0.51 (0.34, 0.76)	0.70 (0.52, 0.95)	<0.01
Adjusted	Ref	0.86 (0.66, 1.13)	0.56 (0.37, 0.85)	0.80 (0.57, 1.12)	0.07
Low birth weight
n	148/2173	135/2100	52/1098	112/1749
Crude	Ref	0.94 (0.75, 1.18)	0.70 (0.51, 0.95)	0.94 (0.79, 1.30)	0.34
Adjusted	Ref	0.90 (0.71, 1.14)	0.65 (0.47, 0.90)	0.86 (0.66, 1.13)	0.15
Small for gestational age
n	195/2096	170/2028	100/1083	136/1707
Crude	Ref	0.90 (0.74, 1.10)	0.99 (0.79, 1,25)	0.86 (0.69, 1.06)	0.19
Adjusted	Ref	0.89 (0.73, 1.09)	0.97 (0.77, 1.24)	0.86 (0.68, 1.08)	0.22
Stillbirth
n	68/2299	75/2231	33/1164	59/1859
Crude	Ref	1.14 (0.82, 1.57)	0.96 (0.64, 1.44)	1.07 (0.76, 1.51)	0.75
Adjusted	Ref	1.05 (0.75, 1.46)	0.83 (0.54, 1.27)	0.83 (0.56, 1.22)	0.31
Neonatal death
n	66/2144	50/2086	21/1100	37/1744
Crude	Ref	0.78 (0.54, 1.12)	0.62 (0.38, 1.01)	0.69 (0.46, 1.03)	0.03
Adjusted	Ref	0.72 (0.50, 1.05)	0.57 (0.35, 0.92)	0.59 (0.38, 0.90)	0.01

¹Q1, ≤0.50 times/d; Q2, >0.50 to 1.00 times/d; Q3, >1.00 to 1.67 times/d; Q4, >1.67 times/d. Q, quartile; Ref, reference.

²Where indicated, adjusted for maternal age, BMI, parity, maternal education level, wealth, food expenditure per day, energy intake, baseline hemoglobin concentration, dietary intake of all other foods, vegetable protein intake, and treatment regimen.

Higher dietary fish intake was associated with lower risk of very preterm birth (medium tertile compared with low; RR: 0.39; 95% CI: 0.28, 0.54; high tertile compared with low; RR: 0.76; 95% CI: 0.58, 0.99; P-trend = 0.02) (Table 3).

TABLE 3.

Univariable and multivariable associations of dietary fish intake with adverse pregnancy outcomes¹

	Tertiles of daily total fish intake, RR (95% CI) 2
Outcome	Low	Medium	High	P-trend
Preterm delivery
n	456/2724	264/2281	373/2313
Crude	Ref	0.69 (0.60, 0.80)	0.96 (0.85, 1.09)	0.41
Adjusted	Ref	0.71 (0.62, 0.82)	0.94 (0.83, 1.08)	0.25
Very preterm delivery
n	152/2724	48/2281	96/2313
Crude	Ref	0.38 (0.27, 0.52)	0.74 (0.58, 0.95)	0.01
Adjusted	Ref	0.39 (0.28, 0.54)	0.76 (0.58, 0.99)	0.02
Low birth weight
n	179/2651	115/2222	153/2247
Crude	Ref	0.77 (0.61, 0.96)	1.01 (0.82, 1.24)	0.97
Adjusted	Ref	0.79 (0.63, 0.99)	1.01 (0.81, 1.26)	0.98
Small for gestational age
n	224/2542	204/2189	173/2183
Crude	Ref	1.06 (0.88, 1.27)	0.90 (0.74, 1.09)	0.30
Adjusted	Ref	1.07 (0.89, 1.28	0.90 (0.74, 1.10)	0.35
Stillbirth
n	89/2813	74/2355	72/2385
Crude	Ref	0.99 (0.93, 1.35)	0.95 (0.70, 1.30)	0.77
Adjusted	Ref	1.01 (0.74, 1.37)	0.91 (0.66, 1.25)	0.57
Neonatal death
n	80/2623	54/2218	40/2233
Crude	Ref	0.80 (0.57, 1.12)	0.59 (0.40, 0.85)	0.01
Adjusted	Ref	0.81 (0.57, 1.14)	0.60 (0.41, 0.88)	0.09

¹Low, 0 times/d; medium, >0 to <0.67 times/d; high, ≥0.67 times/d. Ref, reference.

²Where indicated, adjusted for maternal age, BMI, parity, maternal education level, wealth, food expenditure per day, energy intake, baseline hemoglobin concentration, dietary intake of all other foods, vegetable protein intake, and treatment regimen.

Table 4 summarizes associations of dietary meat and egg intake with poor pregnancy outcomes, comparing women with and without any intake of these foods during pregnancy. Compared with no meat intake, any meat intake was associated with the following: 27% lower risk of preterm birth (RR: 0.73; 95% CI: 0.65, 0.82; P < 0.001), 41% lower risk of very preterm birth (RR: 0.59; 95% CI: 0.46, 0.75; P < 0.001), 36% lower risk of LBW (RR: 0.64; 95% CI: 0.53, 0.78; P < 0.001), 19% higher risk of SGA (RR: 1.19; 95% CI: 1.01, 1.36; P = 0.04), and 31% lower risk of neonatal death (RR: 0.69; 95% CI: 0.49, 0.91; P = 0.01). Compared with no egg intake, any egg intake was protective of very preterm birth (RR: 0.50; 95% CI: 0.31, 0.83; P = 0.01).

TABLE 4.

Univariable and multivariable associations of dietary meat and egg intake with adverse pregnancy outcomes ¹

	Daily total meat intake, RR (95% CI) 2			Daily total egg intake, RR (95% CI) 2
Outcome	No intake	Any intake	P value	No intake	Any intake	P value
Preterm delivery
n	678/3871	415/3447		996/6453	97/865
Crude	Ref	0.69 (0.61, 0.77)	<0.001	Ref	0.73 (0.60, 0.88)	<0.01
Adjusted	Ref	0.73 (0.65, 0.82)	<0.001	Ref	0.83 (0.68, 1.01)	0.07
Very preterm delivery
n	200/3871	96/3447		280/6453	16/865
Crude	Ref	0.54 (0.42, 0.68)	<0.01	Ref	0.43 (0.26, 0.70)	<0.001
Adjusted	Ref	0.59 (0.46, 0.75)	<0.001	Ref	0.50 (0.31, 0.83)	0.01
Low birth weight
n	280/3757	167/3363		399/6276	48/844
Crude	Ref	0.67 (0.55, 0.80)	<0.001	Ref	0.89 (0.67, 1.20)	0.45
Adjusted	Ref	0.64 (0.53, 0.78)	<0.001	Ref	0.89 (0.67, 1.19)	0.43
Small for gestational age
n	295/3622	306/3292		543/6081	58/833
Crude	Ref	1.14 (0.98, 1.33)	0.09	Ref	0.78 (0.60, 1.01)	0.06
Adjusted	Ref	1.19 (1.01, 1.36)	0.04	Ref	0.83 (0.62, 1.07)	0.13
Stillbirth
n	118/3989	117/3564		211/6664	24/889
Crude	Ref	1.11 (0.86, 1.43)	0.42	Ref	0.85 (0.56, 1.29)	0.45
Adjusted	Ref	1.01 (0.77, 1.31)	0.96	Ref	0.77 (0.50, 1.17)	0.23
Neonatal death
n	106/3721	68/3353		157/6233	17/841
Crude	Ref	0.71 (0.53, 0.96)	0.03	Ref	0.97 (0.89, 1.05)	0.41
Adjusted	Ref	0.69 (0.49, 0.91)	0.01	Ref	0.76 (0.46, 1.27)	0.31

¹Ref, reference.

²Where indicated, adjusted for maternal age, BMI, parity, maternal education level, wealth, food expenditure per day, energy intake, baseline hemoglobin concentration, dietary intake of all other foods, vegetable protein intake, and treatment regimen.

When compared with women who had no dairy intake, women with any dairy intake had an 18% reduced risk of preterm birth (RR: 0.82; 95% CI: 0.68, 0.98; P = 0.03) and 47% reduced risk of very preterm birth (RR: 0.53, 95% CI: 0.34, 0.84; P = 0.01) (Table 5).

TABLE 5.

Univariable and multivariable associations of dietary dairy intake with adverse pregnancy outcomes¹

	Daily total dairy intake, RR (95% CI)2
Outcome	No intake	Any intake	P value
Preterm delivery
n	977/6283	116/1035
Crude	Ref	0.72 (0.60, 0.86)	<0.001
Adjusted	Ref	0.82 (0.68, 0.98)	0.03
Very preterm delivery
n	276/6283	20/1035
Crude	Ref	0.44 (0.28, 0.69)	<0.001
Adjusted	Ref	0.53 (0.34, 0.84)	0.01
Low birth weight
n	390/6108	57/1012
Crude	Ref	0.88 (0.67, 1.16)	0.36
Adjusted	Ref	0.87 (0.67, 1.15)	0.33
Small for gestational age
n	520/5919	81/995
Crude	Ref	0.93 (0.74, 1.16)	0.51
Adjusted	Ref	0.98 (0.78, 1.23)	0.87
Stillbirth
n	210/6493	25/1060
Crude	Ref	0.73 (0.48, 1.10)	0.13
Adjusted	Ref	0.67 (0.44, 1.01)	0.06
Neonatal death
n	142/6061	32/1013
Crude	Ref	1.35 (0.92, 1.97)	0.12
Adjusted	Ref	1.29 (0.87, 1.91)	0.21

¹Ref, reference.

²Where indicated, adjusted for maternal age, BMI, parity, maternal education level, wealth, food expenditure per day, energy intake, baseline hemoglobin concentration, dietary intake of all other foods, vegetable protein intake, and treatment regimen.

Discussion

This study examined the relation between the frequency of maternal animal protein intake and preterm birth, very preterm birth, LBW, SGA, stillbirth, and neonatal mortality in a HIV-negative urban cohort in Tanzania. Overall, higher maternal dietary animal protein intake was associated with reduced risk of adverse pregnancy outcomes.

The WHO recommends intake of 0.83 g protein/kg/d to meet the needs of most of the healthy adult population, and an additional 1, 9, and 31 g protein/d for pregnant women in the first, second, and third trimesters, respectively (25). The mean dietary protein intake in this cohort is lower than the RDA for pregnant women (71 g/d in the second half of pregnancy) (26). Our reported protein intake is also lower than the FAO's estimates for the African region at the time of the study (59 g/d) (27). However, these statistics may have been overestimated as they were based on disappearance (i.e., protein that disappears from the food supply rather than protein consumed) (10).

Overall maternal animal protein dietary intake was associated with reduced risk of neonatal death in this cohort. It is noteworthy that whereas total protein was not low in the study, the amount of high-quality protein may be an issue and may lead to deficiencies in critical amino acids for women during pregnancy. Animal protein foods such as fish, meat, eggs, and dairy are important sources of essential amino acids, iron, folate, vitamin B-12, PUFAs such as DHA, and calcium, which play important roles in fetal growth and development (28). Amino acids regulate key metabolic pathways; they also serve as precursors for the synthesis of nitrogenous substances such as NO (28, 29). Low maternal dietary protein intake means that specific amino acids may be deficient in both mother and fetus (29). A deficiency in arginine, for example, may cause preterm labor by stimulating the uterine myometrium due to the reduced bioavailability of NO (28).

Iron, folate, and vitamin B-12 are crucial in erythropoiesis (30). Folate and vitamin B-12 are required for erythroblast proliferation and differentiation whereas iron is an important constituent of hemoglobin, which is essential for the transfer of oxygen to tissues (28, 30). Inadequate prenatal iron intake increases risk of iron deficiency, which may affect fetal growth and development and increase the risk of LBW, preterm delivery, and intrauterine growth retardation (4, 28, 31, 32). Several studies have demonstrated that prenatal iron supplementation reduced the occurrence of preterm birth, LBW, SGA, and neonatal death (32–37).

Previous analysis of nutrient data in this cohort revealed that women with higher animal source iron intake had lower risk of preterm birth, very preterm birth, and neonatal death (20); this pattern of association is similar to what we found for meat intake in this analysis. Therefore, it may be beneficial for nutrition programs in resource-limited settings to promote adequate intake of animal protein foods, particularly to pregnant women if their consumption is low in these settings.

Fish intake was associated with lower risk of very preterm birth in this cohort. There have been conflicting results on the association between dietary fish intake and birth outcomes. A review that included 14 studies assessing the impact of fish intake on birth outcomes found a positive association in 6 studies, an inverse association in 4 studies, and no association in the remaining 4 studies (38). More recently, a British study (39) found no association between prenatal fish intake and preterm birth, whereas studies in Norway (40) and Spain (41) found that seafood intake was protective of preterm birth, LBW, and SGA. In our study, most fish consumed were small dried fish—that is, lean fish that are low in ω-3 fatty acids. Small fish are usually consumed with their bones, contributing to higher calcium intake. However, it is feasible that although fish were consumed frequently, the amounts may have been low and thus not sufficient to contribute to other pregnancy outcomes.

We found that meat intake was associated with lower risk of preterm birth, very preterm birth, LBW, and neonatal death but higher risk of SGA. Our findings are similar to those reported in a European study that found a significant decrease in birth weight with decreased meat intake in late pregnancy (42). Contrary to our findings, a Spanish study found no association between meat intake and SGA (43). It is noteworthy that the studies conducted in developed countries compared diets with more meat consumption and diets with more plant consumption, whereas we compared any meat intake and no meat intake. Furthermore, the level of meat intake is much higher in Europe, so it is possible that there is an inverted U-shaped relation between meat intake and risk of poor pregnancy outcomes. Moreover, in the Tanzanian context, meat intake may be an indicator of a more diverse diet; this may explain why women with any meat intake had lower risk of adverse birth outcomes in our cohort.

Egg intake was protective of very preterm birth in this cohort. Other studies found protective associations between egg intake and SGA (44) and LBW (45). Eggs contain nutrients that may be beneficial during pregnancy as well as during weaning, including vitamin D, folate, iodine, selenium, and long-chain n-3 PUFAs (45, 46). Dietary intake of n-3 PUFAs is associated with reduced recurrence risk of preterm delivery among women with high-risk pregnancies (47). This cohort of women had generally low levels of egg intake. Thus, it may be beneficial to promote prenatal egg consumption in resource-limited settings such as our study context.

In this cohort, dairy intake was protective of preterm birth and very preterm birth. These findings may be explained by the high calcium content in dairy foods. Calcium is a component of bone and acts as a second messenger that regulates numerous biochemical pathways (28). In pregnancy, calcium is important for cell-to-cell adhesion, implantation and placentation, fetal growth, and development (28), and its deficiency is associated with preeclampsia and preterm delivery (28, 48). Moreover, consumption of milk increases blood concentration of insulin-like growth factor 1, which is a major determinant of growth (49). Our findings are supported by other studies that found a protective association between dietary calcium intake and preterm birth (20, 50) and very preterm birth (20). Our findings contrast those from a Chinese study that reported higher odds of preterm birth with frequent dairy consumption; the authors hypothesized that women with frequent consumption of dairy might have considered their diets healthy enough, without paying attention to the context of a balanced diet such as consumption of whole vegetables (51). Other authors have reported a positive association between maternal milk intake and birth weight and fetal growth (52).

The type of benefit accorded by consumption of animal protein foods differs by the type of food consumed and the amino acids and micronutrients that they provide. This may account for differences found in associations of animal protein types with birth outcomes. In this study, we found few associations for SGA and LBW. This may be related to differences in the etiologies for these and other birth outcomes. Previous studies as indicated here have suggested a possible role of caloric intake in the SGA occurrence. However, many studies for LBW have indicated a potential role for iron intake. It is feasible that for other outcomes, consumption of ω-3 fatty acids or calcium intake may be more important. Therefore, there may be different etiologies influencing the different pregnancy outcomes in the study.

Our study limitations include the lack of data on important covariates, such as maternal smoking and alcohol use, which may have affected birth outcomes. However, these practices are generally low in the Tanzanian context, particularly for women. We also lacked data on other socioeconomic factors, such as access to health care. Since mothers were enrolled into the parent trial between 12 and 27 wk of gestation, we lacked data on dietary intake in the first trimester of pregnancy. Additionally, using 24-h recall data may have led to imprecise estimates of dietary intake. Yet, this would likely result in nondifferential exposure misclassification, which would bias results toward the null.

Finally, the data used in this analysis were collected from 2004 to 2006 in Dar es Salaam. Since then, a dietary transition has occurred with increasing consumption of less nutritious foods, particularly due to the availability of cheap ultraprocessed food and beverages in LMICs (53). Moreover, given that the changes favoring an unhealthy dietary pattern and limited physical activity are associated with urbanization, the prevalence of overweight and obesity in urban areas is consistently higher than in rural areas (54, 55). Nevertheless, with the paucity of prenatal dietary data in the region, this analysis provides important information on the potential role of prenatal animal protein intake in the occurrence of adverse pregnancy outcomes, which remain high in the study context.

The major strength of this study was the prospective nature of data collection, allowing for proper temporal relations to be assessed between maternal dietary animal protein intake and birth outcomes. Data collection was performed by trained interviewers and health professionals using standard questionnaires, minimizing errors. Finally, controlling for important confounders in multivariate regression analysis reduced bias in our results. Our findings may be generalizable to other HIV-negative pregnant women in resource-limited settings.

In conclusion, daily animal protein intake among this cohort of pregnant urban Tanzanian women was low when compared with WHO-recommended levels. As such, we determined the following: fish intake was associated with lower risk of preterm birth; meat intake was associated with reduced risk of preterm birth, very preterm birth, LBW, and neonatal death but increased risk of SGA; egg intake was associated with reduced risk of very preterm birth; and dairy intake was associated with reduced risk of preterm birth and very preterm birth. These findings suggest that prenatal counseling and other nutrition interventions for pregnant women should consider consumption of quality animal protein in moderation by women in resource-limited settings as a strategy to address risk of poor pregnancy outcomes.

Notes

The parent trial was supported by a grant from the National Institute of Child Health and Human Development (R01 37701). The funder had no role in the design, analysis, or writing of this article.

Author disclosures: The authors report no conflicts of interest.

Supplemental Table 1 is available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.

Abbreviations used: ASF, animal source food; LBW, low birth weight; LMIC, low- and middle-income country; SGA, small for gestational age; SSA, sub-Saharan Africa.