Contents - Previous - Next

Maternal body mass index and pregnancy outcome in the nutrition collaborative research support program

The Nutrition CRSP
Nutritional status and BMI of the women
Relationship between maternal BMI and other anthropometric variables
BMI and maternal weight gain
BMI and post-partum weight and fat retention
Additional analyses from the Mexico CRSP
Maternal BMI and size of the infant at birth
Maternal BMI and infant size during early lactation
Predictors of maternal BMI
Conclusions and implications for assessment

L. H. Allen1, M. S. Lung'aho1, M. Shaheen2, G. G. Harrison2, C. Neumann2 and A. Kirksey3

1Department of Nutrition, University of California, Davis, CA 95616-8669; 2School of Public Health, University of California, Los Angeles, CA 90024; and 3Department of Foods and Nutrition, Purdue University, W. Lafayette, IN 47907, USA

This paper uses data from the Nutrition CRSP in Egypt, Mexico and Kenya to examine relationships between maternal BMI and pregnancy outcome. Women were studied from the periconceptional period up to 6 months of lactation. No women in Egypt or Mexico, and only two in Kenya where periods of food shortage occurred, had a BMI <18 at conception. Women with a lower BMI in Mexico and Kenya gained more weight and fat in pregnancy and lost more weight and fat during lactation. These counter-intuitive relationships affect the interpretation of weight and body composition measures used to assess energy adequacy during pregnancy and lactation. Lower pre-pregnancy BMI predicted lower infant weights at birth and was a risk factor for low birthweight in Kenya. At 3-6 months post partum, maternal BMI was less strongly related to infant size, and the lean body mass component of BMI appeared to be a more important predictor than fatness.


The Nutrition Collaborative Research Support Program (CRSP) was a cross-country research project funded by the US Agency for International Development between 1981 and 1992*, which investigated relationships between marginal malnutrition and human function, including the outcome of pregnancy. Every possible effort was made to use the same or similar methods across countries. The methodology used in the Nutrition CRSP has been described in detail elsewhere, and the Final Reports are available from each project (Allen et al., 1992; Calloway, Murphy & Beaton, 1992; Kirksey et al., 1992; Neumann, Bwibo & Sigman, 1992). A summary of the main results has been published (Allen, 1993) in addition to many articles. The analyses reported here examine the relationship between maternal body mass index (BMI) and pregnancy outcome. They were performed on the Mexican data at the University of Connecticut, and on the Egyptian and Kenyan data at the University of California, Los Angeles.

* The Egypt CRSP was a collaboration between the Nutrition Institute in Cairo and the University of Arizona, the University of Kansas and Purdue University; principal investigators: Osman Galal, Gail Harrison, Norge Jerome and Avanelle Kirksey. The Kenya CRSP was conducted by the University of Nairobi (Nimrod Bwibo) and the University of California at Los Angeles (Charlotte Neumann). In the Mexico CRSP, the Instituto Nacional de la Nutricion (Adolfo Chavez) collaborated with the University of Connecticut P.I.s, Lindsay Allen and Gretel Pelto. Administrative and scientific oversight, as well as cross-project analysis, were conducted at the University of California, Berkeley by Dons Calloway, Suzanne Murphy and George Beaton.

Correspondence to: Dr L.H. Allen.

The Nutrition CRSP

The Nutrition CRSP in Egypt was located in a pert-urban community near Cairo. Important dietary items included wheat bread, rice, and legumes, supplemented with locally produced cheese, milk, fruits and vegetables. The costs of staples, sugar, meat and oil were subsidized by the government. The households in rural Kenya grew cash crops including coffee and cotton, and some food such as maize, vegetables and legumes. A drought occurred during the project so that energy intake was chronically low for some households and much lower during the drought. The Mexican Nutrition CRSP was based in a rural highland valley, about 3 hours north-west of Mexico City. The economy was subsistence agriculture with manual labour in Mexico City providing an important additional source of income. The main dietary staple was home-produced maize eaten as tortillas which supplied 60% of the dietary energy, and smaller quantities of pasta, beans and rice. Animal products supplied 18% of energy intake in Egypt, 8% in Kenya and 12% in Mexico. In general, food availability and dietary quality were best in Egypt, intermediate in Mexico and poorest in Kenya.

One goal of the Nutrition CRSP was to examine how marginal malnutrition affects the outcome of pregnancy and lactation. To study this question, in each project approximately 100 women were studied from early in pregnancy until 6 months of lactation. Throughout this period maternal food intake and non-breast milk foods consumed by infants were measured for 2 days each month, and maternal and infant anthropometric measurements were made monthly. Morbidity of the mother and infant was recorded from maternal recall every week. Information was collected on the obstetric history of the mother and on various aspects of her household, including demography, socioeconomic status, and food use. Pregnancy outcome measures included birth weight and length, and gestational age measured by the Dubowitz exam.

The major strengths of the Nutrition CRSP are its longitudinal nature, and the fact that similar measures were made across three countries. In addition, information was collected which is used in this paper to test both the biological and social predictors of maternal BMI.

Nutritional status and BMI of the women

As shown in Table 1, the age of the mothers was similar across the three projects, and parity was quite high because many of the households had been selected for participation on the basis of their having children of a specific age, who were also part of the study. The average birth interval was 2-2.5 years, with breastfeeding extending into the second year. Egypt had the highest prevalence of infant mortality and morbidity, because of crowded living conditions and poor sanitation, and also had the lowest rate of maternal literacy. The percentage of low birth weight infants (<2500g) was 4 (4 out of 101) in Egypt, 7 (5 out of 70) in Mexico, and 10 (14 out of 140) in Kenya.

Anthropometric data on the 70-140 women studied during pregnancy are presented in Table 2. Women's heights were similar across the projects, although the Mexican women tended to be slightly shorter. However, the Egyptian women weighed substantially more than women in the other two countries, and were 11 kg heavier than Kenyan women on average. Mexican mothers had intermediate weights. Median BMIs were 25.6 (Egypt), 21.2 (Kenya) and 23.4 (Mexico). These fell within the high (+0.4 Z), low (-0.6Z) and average (+0.1 Z) ranges of US reference women of that age (Abraham, Johnson & Majjar, 1979). A low BMI (<20 during the first trimester) was found in 2% of Egyptian mothers, 32% of the Kenyans and 5% of the Mexicans. No women in the Egyptian or Mexican samples, and only two in the Kenyan sample, had a BMI <18 in the first trimester. A high BMI (>26) occurred in 45% of the Egyptians, 7% of the Kenyans, and 18% of the Mexicans.

Table 1. Maternal characteristics




Age (years)








Birth interval (months)




Maternal literacy (%)




Table 2. Anthropometric data in pregnancy and lactationa




First trimester

Height (cm)

155.3 5.3 (117)

154.3 5.7 (90)

152.5 5.0 (130)

Weight (kg)

61.6 10.4 (96)

50.5 7.7 (90)

54.4 7.2 (74)

BMI (kg/m2)

25.2 4.1 (95)

21.2 2.6 (90)

23.4 3.1 (74)

Weight change (kg)

Trimesters 1- 3

7.8 3.5 (63)

6.4 3.4 (61)

7.7 3.8 (34)

Trimesters 2-3

7.4 3.9 (73)

6.3 3.4 (111)

7.3 3.8 (73)

Lactation 0-6 months

0.2 3.5 (38)

-1.1 3.3 (109)

0.9 3.5 (74)

Birth weight (kg)

3.25 0.61 (101)

3.15 0.40 (122)

3.10 0.40 (70)

Birth length (cm)

50.5 1.3 (53)

49.0 2.3 (116)

49.5 2.2 (60)

Infant weight 6 months (kg)

0.75 1.01 (84)

0.67 1.08 (103)

0.62 1.13 (68)

aValues are mean SD (number in each group in parentheses).
BMI = body mass index.

The BMIs shown in Table 2 reflect the relative weights of the three populations. For example, cross-sectional data on women aged 18-74 years in the same households show that BMI was highest in Egypt and intermediate in Mexico, and increased with age in both countries. In contrast, the BMI of Kenyan women was lower and did not increase as they became older. Based on these data it has been suggested that the degree of change in women's BMI with age might be a useful indicator of general energy adequacy in a population group (Calloway et al., 1988).

The pattern and amount of pregnancy weight gain were surprisingly similar across the three locations (Table 2). The average gain was ~7 kg, which is only half that recommended in the USA (Institute of Medicine, 1990). The pattern of weight change through pregnancy was similar across the projects, but only the Egyptian group avoided weight loss between 1 and 6 months post partum. This may be another useful indicator of the general energy adequacy of a population. In general the weight of the Egyptian and Mexican women at 6 months post partum was higher than at conception, while in Kenya weights were similar to those early in pregnancy.

Table 3. Preconception anthropometry inter-correlationsa,b




















- 0.03
















Fat mass










% fat










Arm muscle







Lean mass










aSubscapular and suprailiac skinfolds were difficult to obtain in Egypt because long garments were worn. Fatness was calculated from only triceps and biceps measures. In Kenya only the sum of 4 skinfolds was reported so the correlations were not run with single-site skinfold measures.

bAll correlations > 0.27 were statistically significant (P < 0.05).
Wt = weight; Ht = height; BMI = body mass index.

Relationship between maternal BMI and other anthropometric variables

In addition to monthly weight and height measures, skinfold thickness was recorded at four sites (triceps, biceps, subscapular and suprailiac), and mid-upper arm circumference was measured. Correlations among these measures were examined in order to understand better how each was related to estimates of fatness and lean body mass. Body fat and lean masses were calculated from the formula of Durnin & Womersley (1974) which linked skinfold measures to density to measure both fat and lean mass (Sir), 1961). Mid-upper arm muscle area was estimated from mid-upper arm circumference and triceps skinfold thickness (Frisancho, 1981).

BMI was strongly correlated with weight in all three countries but was poorly related to height (Table 3). BMI reflected fatness, based on its strong association with skinfold thickness at all four sites (biceps and suprailiac correlations not shown), and with fat mass and the percentage of body weight as fat. It was also quite strongly correlated with arm muscle mass and lean mass. These correlations indicate that women with a higher BMI have more fat and lean tissue, and a higher percentage of fat, but are not taller. Mid-upper arm circumference was also tested in these correlations but was dropped from subsequent analyses because it was associated equally well with both fat and lean tissue, and was therefore difficult to interpret.

Contents - Previous - Next