Previous - Next
In both Mexico and Kenya total pregnancy weight gain was strongly negatively correlated (Spearman's correlations) with maternal weight, BMI, and triceps skinfold thickness (Table 4). Other indicators of fatness (subscapular skinfold, fat mass and % fat) were also strongly negatively associated with weight gain in Mexico and, less strongly so, in Kenya. The negative correlation between BMI and weight gain was strongest in the first trimester in Kenya (r = -0.38, P < 0.001) and in the second trimester in Mexico (r = -0.25, P < 0.05). A similar pattern was seen in these countries when weight gain was expressed as a percent of periconceptional weight (Table 4). No such relationships were seen in Egypt. However, Egyptian mothers with a BMI >27.4 gained only 6.0 ± 2.3 kg in their last two trimesters, compared to 8.2 ± 3.4 kg for women with a BMI between 22.6 and 27.4. In no case did maternal height predict pregnancy weight gain. The negative associations with gestational weight gain were strongest between triceps skinfold thickness and measures of fatness, although lean body mass was also a negative predictor of gain.
Table 4. Spearman's correlations of periconceptional anthropometry with the overall weight gain during pregnancya,b
|
Maternal characteristic |
Egypt |
Kenya |
Mexico | |||
|
Total |
% initial |
Total |
% initial |
Total |
% initial | |
|
Weight |
0.06 |
- 0.18 |
- 0.44 |
- 0.42 |
- 0.31 |
- 0.46 |
|
Height |
0.07 |
-0.13 |
0.13 |
0.15 |
0.05 |
0.12 |
|
Body mass index |
0.06 |
-0.16 |
-0.46 |
-0.44 |
-0.34 |
-0.48 |
|
Triceps |
0.02 |
-0.12 |
-0.37 |
-0.42 |
-0.55 |
-0.64 |
|
Subscapular |
-0.25 |
-0.28 |
-0.33 |
-0.46 | ||
|
Fat mass |
0.13 |
-0.08 |
-0.27 |
-0.31 |
-0.33 |
-0.48 |
|
% fat |
0.08 |
-0.02 |
- 0.18 |
-0.21 |
-0.31 |
- 0.45 |
|
Arm muscle |
0.13 |
0.02 |
-0.04 |
-0.03 | ||
|
Lean mass |
0.13 |
- 0.06 |
- 0.27 |
- 0.31 |
- 0.23 |
- 0.27 |
aThe initial measure of maternal anthropometry measure was taken during the interval between 2 months prior to conception and 42 days of pregnancy.
bAll correlations > 0.25 were statistically significant (P < 0.05).
An inverse relationship between maternal weight (r = -0.15, n = 125) or BMI (r = -0.27) and pregnancy weight gain was also reported for Taiwanese women by Adair, Pollitt & Mueller (1983). The average BMI of the Taiwanese women was 20, or slightly lower than the Kenyan group, although they appeared to have adequate energy intakes. In addition, those Taiwanese women with higher triceps and subscapular skinfold thickness prior to pregnancy lost substantially more fat from these regions during pregnancy. For example, the correlation between prepregnancy triceps skinfold thickness and change in this measure during pregnancy was -0.42. Similarly, in an older study of well-nourished women in Britain, Taggart et al. (1967) observed that those with a greater skinfold thickness at 10 weeks of pregnancy had less of an increase, and often a loss, in skinfold thickness by 38 weeks of gestation. This negative relationship between maternal BMI and pregnancy weight gain is evident in women in wealthier countries as well, so the most recent pregnancy weight gain recommendations in the USA are based on maternal BMI in early pregnancy (Institute of Medicine, 1990). For example, women with a high BMI (>26.0) are recommended to gain between 7 and 11.5 kg, compared to a 12.5-18 kg recommendation for thin women (BMI <19.8). These recommendations are based on an analysis of 1980 data collected in the National Perinatal Collaborative Study, which showed that women with a higher BMI gained substantially less weight but that this was still compatible with a healthy pregnancy outcome. On the other hand, thinner women are urged to gain more weight to reduce the risk of lower birth weight.
In Kenya the women were relatively thin and in the period of hunger and energy shortage they gained less weight during pregnancy. Nevertheless it is important to note that thinner Kenyan women still gained more than those who were fatter. This suggests that the inverse relationship between maternal BMI and pregnancy weight gain is driven through an effect of body composition, rather than current energy intake.
In all three locations women who were heavier and fatter (BMI, fat mass, and % fat) at conception had retained substantially less weight and fat at 2-4 weeks post partum, reflecting the lower weight gain of fatter women during pregnancy. For example, correlations between early pregnancy BMI and early post partum retention (as a percentage of early pregnancy weight) were -0.84 in Egypt, -0.63 in Kenya and -0.52 in Mexico. Correlations with % body fat in early pregnancy and retained weight post partum were -0.90, -0.50 and -0.55 in the three countries, respectively. Initial fatness was a much stronger predictor of lower post-partum retention than was lean mass. A similar pattern was found in the Taiwanese sample described by Adair et al. (1983); those who were fatter preconception gained less weight in pregnancy, and lost more weight and fat in lactation.
The Mexico data set has been used to explore the effect of maternal fatness on subsequent anthropometric changes in somewhat more detail (Lung'aho, 1992). This data set is relatively unusual because longitudinal data were available for 36 women starting prior to conception, or at the latest within 42 days of conception, continuing throughout pregnancy until 8 months of lactation. The women were grouped into terciles based on their BMI at conception. These were called the low, moderate and high BMI terciles, but all BMIs were within the normal range. Specifically, the lowest tercile ranged from 18.5 to 21.9, the moderate tercile from 22.0 to 21.9, and the high tercile from 24.0 to 26.5. None of the women shifted from one tercile to another during the study.
Figures 1 and 2 illustrate the changes from preconception values for weight and triceps skinfold thickness during pregnancy and lactation. Within this group of 36 women, as for the larger sample described above, those in the lowest BMI tercile at conception gained cumulatively more weight throughout pregnancy, especially in the first trimester - a period that is not often measured. Subsequently, these thinner women had more retained weight post partum,
Changes in skinfold fatness mirrored the weight changes. For example, the thinnest group had a small increment in triceps skinfold thickness early in pregnancy which was retained during lactation. In contrast, women in the highest BMI tercile lost substantial amounts of fat from the triceps area throughout pregnancy and early lactation, and by 6 months of lactation this deficit was still apparent. The pattern was similar for subscapular skinfolds (not shown), with the lowest BMI group gaining the most fat in this region. On average, the highest BMI group failed to gain subscapular fat during pregnancy, but did gain it in this location very early in lactation.
These analyses indicate that a lower BMI at conception predicts higher weight gain during pregnancy, as well as more fat gain on the extremities and trunk. In contrast, fatter women actually lose fat from the triceps site throughout pregnancy and lactation and fail to gain any on the trunk until lactation. Subsequently, at 6 months post partum there is a tendency for the fatter women to have less fat on the arm than at conception, but more on the trunk.
Maternal energy intake paralleled these gains and losses in weight and fat. In the lowest BMI group energy intake was higher at conception (although this was not a statistically significant difference) as well as during every trimester of pregnancy. We cannot rule out the possibility that the low BMI group was the most physically active. The correlation between median energy intake and weight gain during pregnancy was 0.41 in trimester 2, 0.30 in trimester 3, and 0.32 overall (P < 0.05). Thus, the higher weight gain of the lowest BMI group was accompanied by a higher energy intake; the correlation between BMI at conception and average energy intake during pregnancy was -0.45 (P < 0.01), while between BMI and energy intake/kg it was -0.69 (P < 0.001). In Egypt and Kenya, women with a higher BMI also consumed less energy during pregnancy; the correlations between BMI and average energy intake/kg in pregnancy were -0.71 (P < 0.001) in Egypt and -0.36 (P < 0.01) in Kenya. In lactation, the highest BMI group in Mexico, which had been losing fat prior to conception, started to regain their fat (Fig. 2). This was accompanied by a dramatic increase in energy intake so that it became higher than in the other two BMI groups; median daily intakes between 6 and 8 months of lactation were 6.59 MJ (2761 kcal) in the lowest BMI tercile, 6.74 MJ (2321 kcal) in the intermediate and 7.58 MJ (3177 kcal) in the highest.
It is well established that women with a higher BMI give birth to heavier infants. This was also true for the Nutrition CRSP participants in
Mexico and Kenya. The correlation between maternal BMI in the first trimester and birth weight was 0.29 (P < 0.05) in Mexico and 0.59 (P < 0.001) in Kenya. All of the low birth weight infants in Kenya were born to women with a BMI <21. BMI was also correlated with birth length in Kenya (r = 0.57, P < 0.001) but not in Mexico. The Egyptian data were not analysed because of the small number of women who had both early pregnancy and birth measures. The mechanism that explains why there is a strong positive association between BMI and birth weight is not understood, but the inverse relationship between energy intake and BMI described for Mexican women in the preceding section suggests that it may not be mediated through food intake.
Data from the National Natality Survey in the USA have been used to illustrate the existence of an interaction between pregnancy weight gain, birth weight and maternal BMI (Institute of Medicine, 1990). Women with a low BMI (<19.8) early in pregnancy tend to produce the lowest birth weight infants, but this effect can be partially overcome by higher weight gain. In contrast, very overweight women (BMI >29) produce the highest birth weight infants but gain relatively little weight in pregnancy, so that pregnancy weight gain has relatively little influence on birth weight in this group. Ideal weight, and moderately overweight groups are intermediate in these responses. Overall, the USA data illustrate that the thinnest women with the lowest pregnancy weight gain are at greatest risk of having a low birth weight infant.
The highest prevalence of low birth weight infants (<2500g) in the Nutrition CRSP was found in the Kenyan sample. Here 14 out of 140 infants were classified as low birth weight. The influence of maternal prepregnancy weight was very strong (BMI was not used in these Kenyan analyses because it was so strongly correlated with weight). Women in the lowest weight quartile (<45 kg), in either trimester 1 or 2, had an odds ratio of 9.1 for low birth weight (Ngare, 1990). Similarly, being below the 25th percentile for mid-upper arm circumference (MUAC <23.7 cm) in trimester 1 increased the odds ratio for low birth weight to 3.5, while in trimester 2 a MUAC <25 raised the odds ratio to 12.6. A logistic regression analysis to predict the risk of low birth weight showed that maternal BMI, parity, maternal hemoglobin, and socioeconomic status correctly classified 86% of the cases. During the period of greatest food shortage in Kenya the percentage of infants weighing between 2500 and 2800 g at birth increased from 16% to 28%; again, while BMI has a major impact on birth weight, inadequate energy intake and weight gain can modulate this influence.
Maternal BMI post partum (which was little different from BMI early in pregnancy) was a less strong predictor of infant weight during the first 6 months post partum. Maternal weight and BMI post partum retained a positive relationship with infant weight at 1 month; r = 0.23 in Egypt, 0.40 in Kenya and 0.42 in Mexico (all P < 0.05). This relationship became less strong at 3 and 6 months (r = 0.29 in Kenya, 0.20 in Mexico) and actually became negative in Egypt (r = -0.33). The relationships between periconceptional maternal fat mass, percent fat and infant weight seen in the Mexico and Kenya groups at 1 month certainly disappeared over time, so that maternal skinfolds, fat mass and % fat at 1 month post partum were unrelated to infant size at 3 or 6 months in any of the three projects. However, there were significant relationships between infant weight at 3 or 6 months and maternal lean mass; at 3 months the correlation was r = 0.24 in Kenya and 0.33 in Mexico (P < 0.05).
In all three countries, maternal BMI was positively related to infant length at 3-6 months (correlations 0.26-0.37). Associations between maternal lean body mass and length at 6 months were significant for Kenyan (r = 0.25, P < 0.05) and Mexican (r = 0.42, P < 0.001) infants, but no relationships were seen with any measures of fatness.
These analyses suggest that maternal lean mass is a stronger predictor of infant weight and length at 3 and 6 months than is maternal fatness. Similar to the situation in most developing countries, growth-faltering of infants started soon after birth in all three projects. The importance of maternal size and body composition in this phenomenon should be a research priority.
Because of the extensive amount of information collected in the Nutrition CRSP, it was possible to explore predictors of maternal BMI. BMI was unrelated to socio-economic status, the value of the family's house, or the education of the mother in any of the projects, including Kenya where energy intake was inadequate. Age had a positive influence only in Egypt, while parity had no relationship to BMI in any location. Finally, energy intake was strongly negatively correlated with BMI in Egypt (with energy intake/kg, r = -0.71, P < 0.01); perhaps fatter women expended less energy in physical activity. The same pattern occurred in Mexico (r = -0.63, P < 0.001). In Kenya, where energy intake was inadequate for some individuals, BMI was not significantly related to the mother's energy intake (r = -0.05).
In the Nutrition CRSP there was a gradation across the projects in terms of energy adequacy, assessed by BMI as well as by other anthropometric measures and energy intake. Energy availability was highest in Egypt followed by Mexico, and generally adequate in both locations. In spite of the severe food shortage that occurred Kenya, low values for BMI (<18) were only observed in two of the Kenyan women.
Even though there were not many cases of low BMI, it is clear that BMI within the normal range influences pregnancy outcome. Specifically, in Mexico and Kenya, as in wealthier populations, women with a lower BMI at conception gained more weight and fat in pregnancy and lost more weight and fat during lactation. In contrast, fatter women lose fat in pregnancy and regain it in lactation. These results clearly have implications for the use of weight gain in pregnancy and weight loss in lactation as tools for assessing current dietary energy adequacy. The weight changes are the reverse of what would be intuitively expected, if BMI is accepted as a marker not only of nutritional status but of food availability. These data also emphasize the importance of considering BMI when interpreting skinfold thickness and body composition changes associated with pregnancy and lactation.
In the CRSP, as in many other studies, low maternal BMI in early pregnancy was associated with lower birth weight in all three projects. It is probable that a combination of low maternal BMI and low weight gain presents the highest risk of low birth weight, and that both measures are needed to best predict this risk. BMI also predicted birth length in Egypt and Kenya. Post partum, BMI remained positively associated with infant weight at 3 months in Mexico, and at 3 and 6 months in Kenya. Also, post-partum BMI had a fairly strong positive relationship to infant length throughout the 6 month period although the strength of this association weakened with time. The relationship between maternal BMI and infant length appeared to reflect lean and fat mass and was not associated with skinfold thickness. In view of the pervasive early growth-faltering that occurs in infants in developing countries, the role of maternal size and body composition in this phenomenon deserves further attention.
Acknowledgements - The Nutrition CRSP was supported by grants #DAN- 1309-G-SS- 1070-00 and DAN- 1309-A 00-9090-00 from the US Agency for International Development.
Abraham S. Johnson CL & Majjar ME (1979): Weight and height of adults 18-74 years of age, United States, 1971-1974. Vital Health Statistics Series 11, No. 211. US Washington, DC: Government Printing Office.
Adair LS, Pollitt E & Mueller WH (1983): Maternal anthropometric changes during pregnancy and lactation in a rural Taiwanese population. Hum. Biol. 55, 771-787.
Allen LH (1993): The Nutrition CRSP: what is marginal malnutrition, and does it affect human function? Nutr Rev. 51, 255-267.
Allen LH, Backstrand JR, Chávez A & Pelto GH (1992): People cannot live by tortillas alone. The results of the Mexico Nutrition CRSP. Storrs, CT: Department of Nutritional Sciences, University of Connecticut.
Calloway DH, Murphy SP & Beaton GH (1988): Food intake and human function: a cross-project perspective. Berkeley, CA: Department of Nutritional Sciences, University of California.
Calloway DH, Murphy S. Balderston J. Receveur O. Lein D & Hudes M (1992): Village nutrition in Egypt, Kenya and Mexico: looking across the CRSP projects. Berkeley, CA: Department of Nutritional Sciences, University of California.
Durnin JVGA & Womersley J (1974): Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years. Br J. Nutr 32, 77-97.
Frisancho R (1981): New norms of upper limb fat areas for assessment of nutritional status. Am. J. Clin. Nutr 34, 2540-2545.
Institute of Medicine (1990): Nutrition during pregnancy. Washington DC: National Academy of Sciences Press.
Kirksey A, Harrison GG, Galal OM, McCabe GP, Wachs TD & Rahmanifar A (1992): The human costs of moderate malnutrition in an Egyptian village. West Lafayette, IN: Department of Foods and Nutrition, Purdue University.
Lung'aho, MS (1992): Importance of body composition in pregnant and lactating Mexican women. Storrs, CT: University of Connecticut, Department of Nutritional Sciences.
Neumann C, Bwibo N & Sigman M (1992): Diet quantity and quality. Functional effects on rural Kenyan families. Los Angeles, CA: School of Public Health, University of California.
Ngare DK (1990): Simple predictors of low birthweight at the community level: their potential application in primary health care for the prevention of adverse pregnancy outcomes in rural Kenya. Ph.D. dissertation. Los Angeles, CA: University of California School of Public Health.
Siri WE (1961): Body composition from body fluid spaces and density: analysis of methods. In Techniques for measuring body composition, pp. 223-244. Washington DC: National Academy of Sciences, National Research Council.
Taggart NR, Holliday RM, Billewicz WZ, Hytten FE & Thomson AM (1967): Changes in skinfolds during pregnancy. Br J. Nutr 21, 439-451.
Kennedy: I was interested in the data where you stratified BMI by the age of the women. In Kenya it was very flat with increasing age, and we found similar results with our Kenya data. In fact, most African rural data show the same thing but in urban populations BMI does go up with age. Perhaps this is a good indicator of overall energy availability? We don't find the same effect in our Asian data sets even though absolute BMIs tend to be lower.
Ferro-Luzzi: You have shown in Kenya a food shortage where people may have to starve for a whole day. Did you find any effect on BMI or reproductive performance?
Allen: Yes, there was a loss of 2-3 kg during the drought period. BMI and activity went down and there was a doubling of the incidence of birth weights between 2.5 and 2.8 kg.
Ferro-Luzzi: Do you have a breakdown of micronutrient deficiency by BMI class?
Allen: No, I don't, but it would be interesting. Prentice: When I see these low BMIs catching up during pregnancy it looks like regression towards the mean. I would like to be reassured that that is not part of the problem. Do we have any data on non-pregnant non-lactating women, or on men, looked at again some years later which would indicate the size of this effect?
James: You infer that the greater the parity of the women, the greater the proportion of low birth weight children.
Allen: Yes, this is true in Kenya, but there are very few primiparous women in the sample.
Mascie-Taylor: How much of the low haemoglobin count is diet related and how much due to parasites?
Allen: Both are involved. In a highland region of Mexico there may be subclinical parasitization but because of low dietary iron availability the intake is only half of that required.