Previous - Next
Introduction
Socioeconomic disparities in IUGR: Mediating factors
Secular trends in IUGR and its determinants
Are there residual socioeconomic disparities in IUGR?
References
Discussion
MS Kramer
Correspondence: Dr MS Kramer
Departments of Pediatrics and of Epidemiology and Biostatistics, McGill University, Faculty of Medicine, 1020 Pine Avenue West, Montreal, Quebec H3A 1A2, Canada
This paper reviews the evidence bearing on socioeconomic determinants of intrauterine growth retardation (IUGR). The primary focus is on those factors with a quantitatively important impact from a public health perspective, as indicated by their large etiologic fraction (population attributable risk). In developed countries in which a sizeable proportion of women smoke during pregnancy, cigarette smoking is associated with the largest etiologic fraction (by far), followed by low gestational weight gain (primarily due to low energy intake) and low pre-pregnancy body mass index (BMI). In developing countries where undernutrition is prevalent and pregnant women do not smoke, low maternal weight gain and BMI assume even greater importance, as does short maternal stature.
A major section of the paper
concerns the large within-country socioeconomic disparities in IUGR and the possible
mechanisms underlying these disparities. In developed countries, differences in cigarette
smoking explain a large part of the disparity; low weight gain and short stature may also
be important mediators in some settings. Future etiologic studies should assess a wide
scope of potential determinants and will require large sample sizes to control for their
mutually confounding effects.
One of the most robust findings in epidemiologic research on the etiology of low birth weight (LBW) has been the large socioeconomic disparities in both intrauterine growth retardation (IUGR) and preterm birth (Kramer, 1987; Parker et al, 1994; Wilcox MA et al, 1995; Kogan, 1995). In a now classic paper, Villar and Belizan reported data from 11 different regions in developed countries and 25 in developing countries (Villar and Belizan, 1982). Their analysis indicated that in developing countries, most low birth weight is due to IUGR, whereas in developed countries (especially those with the lowest LBW rates), most is due to preterm birth (Villar and Belizan, 1982). Most of the differences in LBW rates between developed and developing countries appear attributable to an increased prevalence of IUGR, rather than preterm birth, with relative risks of 6.6 and 2.0, respectively (Villar and Belizan, 1982).
Even within developed countries,
however, socioeconomic disparities in birth weight have been consistently reported. These
disparities occur not only in countries like the United States (Parker et al,
1994), with vast differences between the rich and the poor, but even in more 'socialistic'
countries like Canada (Wilkins et al, 1991), Sweden (Ericson et al, 1993),
and Denmark (Olsen and Frische, 1993) with less extreme contrasts of wealth and poverty
and with universal access to high-quality prenatal and other medical care. In multiracial
countries, IUGR rates have been shown to differ considerably by race, with increased IUGR
rates among blacks in the U.S. (Horon et al, 1983; Showstack et al, 1984;
Linn et al, 1983), aboriginals in Australia (Seward and Stanley, 1981; Sayers and
Powers, 1993), and Asians in the U.K. (Dawson and Golder, 1982; Moore et al, 1995).
Although a small portion of the difference may be due to true biological differences
between the racial groups (Wen et al, 1995), much of it is probably linked to
socioeconomic disadvantage among these racial groups. Interestingly, some other minority
groups do not have the same apparent problem with IUGR. These include American Indian
groups in Canada (Munroe et al, 1984; Thomson, 1990), North African immigrants in
Israel (Yudkin et al, 1983), and Mexican immigrants in the U.S. (Guendelman and
English, 1995); despite their socioeconomic disadvantage relative to the predominant
racial/ethnic groups, these groups show birth weight distributions shifted somewhat to the
right relative to the distributions dominant in their respective countries, with
correspondingly low rates of IUGR.
Having more money or education probably has no direct effect on how fast a fetus grows in utero. In other words, socioeconomic disadvantage is probably not a direct, independent determinant of fetal growth. Rather, socioeconomic disadvantage may lead to adverse psychological, behavioral, or other environmental exposures that restrict fetal growth. It is important to understand these direct-acting etiologic determinants, both in terms of their relative risks and their etiologic fractions. The relative risk (RR) indicates how much more likely the adverse outcome (IUGR in this case) occurs in women with vs those without a given risk factor; the RR is a measure of the strength of association between the risk factor and IUGR. The etiologic fraction (also know as the population attributable risk or attributable fraction) reflects both the relative risk of a given risk factor and its prevalence in the population of interest. It is defined as the proportion of IUGR occurring in a population that can be attributed to the risk factor or, equivalently, the proportion of IUGR in the population that would theoretically disappear if exposure to the risk factor were eliminated. The etiologic fraction (EF) can be estimated as follows:
where P is the prevalence of the risk factor and RR its corresponding relative risk for IUGR (Levin, 1953). Thus the relative risk compares the risk in individual women with and without the risk factor, while the etiologic fraction reflects the population burden of IUGR due to the risk factor.
Much is known about the attributable causes of IUGR, which are depicted schematically in the Figure. In this pie diagram, the size of the pie slice corresponding to each determinant is a rough indication of its importance (in terms of etiologic fraction) relative to the other determinants shown in the pie. It is an updated version originally based on a meta-analysis of the English- and French-language literature published between 1970 and 1984 (Kramer, 1987). The diagram should be interpreted in relative terms only; because the determinants shown are not mutually exclusive, any given woman may have more than one of them. Thus, the total etiologic fraction of all the determinants of IUGR is less than the 100% suggested by the diagram (Bruzzi et al, 1985).
The Figure represents a developed country setting in which approximately 30% of the women smoke cigarettes during pregnancy. In this type of setting, the risk factors with the highest relative risks are pre-eclampsia and cigarette smoking. Because cigarette smoking is so much more prevalent than pre-eclampsia, however, cigarette smoking is associated with a far higher etiologic fraction, i.e., its corresponding pie slice is much larger. Other factors with high etiologic fractions (representing somewhat lower relative risks but high prevalences) include low gestational weight gain (partially, but not exclusively, reflecting low energy intake), low prepregnancy body mass index (BMI), primiparity, short maternal stature, and nonwhite racial/ethnic origin. High levels of alcohol consumption (two or more drinks per day) (Zuckerman and Hingson, 1986; Walpole et al, 1990) and the use of marijuana or cocaine (Hatch and Bracken, 1986; Zuckerman et al, 1989; Shiono et al, 1995) may also restrict fetal growth, but their etiologic fractions are quite low, owing to low prevalences of exposure. The effect of high caffeine consumption (> 300 mg/day) is more controversial; although such consumption is highly prevalent, its corresponding relative risk is only modestly (if at all) elevated (Martin and Bracken, 1987; Dlugosz and Bracken, 1992; Fortier et al, 1993).
In a developing country setting in which pregnant women do not smoke cigarettes, the pie diagram would look quite different. The 'slice' corresponding to cigarette smoking would disappear, while the slices corresponding to low weight gain, low prepregnancy BMI, and short stature would increase in size considerably. In addition, malaria would occupy a substantial slice in those settings where the disease is endemic.
For a risk factor to be an important mediating variable between socioeconomic disadvantage and IUGR, it must satisfy two criteria. First, it must have a high etiologic fraction for IUGR. Second, it must be strongly associated with socioeconomic status, i.e., it should be far more prevalent in low-SES population groups. The main candidates for such mediating variables of IUGR include maternal nutritional factors (maternal stature, prepregnancy body mass index, and energy balance, i.e., gestational weight gain); cigarette smoking; alcohol consumption; narcotic, cocaine, marijuana, and other drug exposure; maternal work during pregnancy; and stress, anxiety, and other psychological factors. A review of the literature of the evidence concerning maternal nutritional determinants of IUGR is the subject of another paper in this supplement. My mandate is to focus on the evidence concerning the non-nutritional factors. Nonetheless, I will begin by briefly reviewing the evidence concerning the relationship between maternal nutrition and socioeconomic disparities in IUGR.
Low prepregnancy BMI is certainly not an explanation for socioeconomic disparities in IUGR in developed country settings, since BMI is inversely associated with socioeconomic status (in other words, BMIs are higher in women of low socioeconomic status) (Flegal et al, 1988). Low BMI may be an important factor in developing countries, however, at least in those developing countries where maternal obesity is less of a problem than it is in most industrialized countries. In countries with high prevalences of maternal undernutrition, it is reasonable to presume that undernutrition is more prevalent among those from unfavorable socioeconomic backgrounds, and thus low prepregnancy BMI could be partly responsible for high IUGR rates seen among the poor in many developing countries. Maternal stature is known to be associated with socioeconomic status, although the direction of cause-and-effect is far from certain (Thomson and Billewicz, 1963). Nonetheless, in developed country settings, differences in stature according to socioeconomic status are small and, given the modest etiologic fraction for maternal stature, unlikely to explain a sizeable portion of socioeconomic disparities in IUGR. Gestational weight gain has also been reported to vary by SES (Taffel, 1986). Low weight gains appear to be more common in women with low educational attainment, while high weight gains are more prevalent among the highly educated (Taffel, 1986). Given the large etiologic fraction associated with low gestational weight gain, this may well represent an important mechanism for the high rate of IUGR among the socioeconomically disadvantaged.
Smoking is probably the most important variable mediating socioeconomic disparities in IUGR in developed country settings. In most such settings, cigarette smoking is both far more common and heavier (i.e., larger number of cigarettes smoked per day) in women from lower socioeconomic backgrounds (Subcommittee on Nutritional Status and Weight Gain During Pregnancy et al, 1990; Hagland et al, 1993). Moreover, as the adverse effects of smoking in pregnancy have become widely recognized by physicians and the general public, the socioeconomic gradient in cigarette smoking has widened (Subcommittee on Nutritional Status and Weight Gain During Pregnancy et al, 1990).
Studies of strenuous work and physical activity during pregnancy have focused on preterm birth (Mamelle et al, 1984; Saurel-Cubizolles and Kaminski, 1987; Homer et al, 1990; Henriksen et al, 1995). Although most epidemiologic studies have not found that prolonged standing or strenuous activity at work is associated with IUGR, there are a few exceptions (Launer et al, 1990; Henriksen et al, 1995). Because a very small fraction of women are exposed to prolonged standing or walking at work, however, its etiologic fraction is likely to be extremely low, i.e., it is not an important mediating variable for socioeconomic disparities in IUGR. Vigorous aerobic exercise that continues into the third trimester may be associated with a slight increase in IUGR (Clapp and Dickstein, 1984), although here too, the small number of women engaging in vigorous exercise during this critical period of fetal growth strongly suggests that it is not an important mediating variable from a population perspective.
Finally, epidemiologists have become
increasingly interested in stress, anxiety, and other psychological factors as risk
factors for adverse pregnancy outcomes. Although most of the evidence for the adverse
effects of these psychological factors bears on preterm birth (Nuckolls et al,
1972; Newton and Hunt, 1984; Lou et al, 1992; Lobel et al, 1992; Nordentoft et
al, 1996) and suggests no effect on fetal growth (Nordentoft et al, 1996; Aarts
and Vingerhoets, 1993; Hedegaard et al, 1996), one study suggests a possible
association with IUGR in thin women (Cliver et al, 1992). An overview
(meta-analysis) of clinical controlled trials shows little if any benefit of social
support on IUGR, with a pooled odds ratio of 1.06 and a 95% confidence interval of
0.86-1.30 (Hodnett, 1994). None of these trials, however, has demonstrated the
effectiveness of the social support in correcting or reducing the stress, anxiety, or
depression of the women who are offered the intervention.
Both Canada and the United States have witnessed a modest decline in the incidence of low birth weight over the last two decades. Most or all of this decline is due to a lower incidence of IUGR, rather than of preterm birth (Ng and Wilkins, 1994; Kessel et al, 1984; Division of Nutrition, National Center for Chronic Disease Prevention and Health Promotion, 1994). In Canada, the birth weight of term infants increased approximately 130-140 grams between 1972 and 1986, with no apparent change in the gestational age distribution or in the growth in infants delivered preterm (Arbuckle and Sherman, 1989). Thus in both the United States and Canada, the only important change in fetal growth over the last several decades has been an increase in the size of term infants.
Based on simultaneous trends in
etiologic determinants, this increase in the size of term births appears to reflect three
primary factors: (1) a secular trend toward increasing prepregnancy body mass index
(Flegal et al, 1988), reflecting a general trend toward increasing adiposity (which
has been particularly evident in North America, but observed in most western countries);
(2) a marked secular trend toward higher gestational weight gains, which have increased
from 9 or 10 kilograms in the 1950's and 60's to 15 kilograms, on average, in recent years
(Taffel; Eastman and Jackson, 1968; Niswander and Jackson, 1974; Keppel and Taffel, 1993);
and (3) substantial decreases in cigarette smoking among pregnant women (Kleinman and
Kopstein, 1987), presumably in response to increasing knowledge and publicity about the
adverse effects of smoking on health, in general, and on pregnancy outcome, in particular.
In an earlier section, I reviewed the evidence concerning factors that may mediate the socioeconomic disparities in IUGR incidence. Leading candidates for quantitatively important (from a population perspective) factors include cigarette smoking (both frequency and severity), low gestational weight gain (presumably due to low energy intake), and short stature. Alcohol, drugs of abuse, and maternal work and physical activity may explain a small, additional portion of the disparity. But can the entire socioeconomic gradient in IUGR be explained by the above-mentioned risk factors? This is an important question for both perinatal epidemiologists and public health policy makers. Unfortunately, the answer is unknown. Most research in this area has tended to focus on one or two factors at a time. Few have adequately measured all of these potentially mediating variables, and most have had insufficient sample sizes to do so. Thus it is unknown whether any residual socioeconomic disparities remain after accounting (controlling) for known, or strongly suspected, mediating variables. This is a promising area for future research that may not only uncover new etiologic determinants of IUGR, but also help reduce its high incidence among the socially disadvantaged.