Ethnic differences in maternal birthplace and marital status may explain most of the established association between Arab ethnicity and a reduced risk of preterm birth, according to a Michigan study of all births over the period 2000–2005.1 The prevalence of preterm births was slightly lower among Arab Americans than among non-Arab whites (8% vs. 9%), and the risk of such births was reduced for the former group when each of six background characteristics was considered (odds ratios, 0.7–0.9). However, when all covariates were analyzed simultaneously, no association was found between preterm births and Arab ancestry.
Arab Americans make up a greater share of Michigan's population than of any other state's population. To assess what factors contribute to the lower risk of preterm birth among Arab Americans, researchers analyzed data on all 617,451 births that occurred in Michigan between September 2000 and March 2005, as reported by the state's Department of Community Health. The main outcome was preterm births, defined as births at less than 37 weeks of gestation. Self-reported race and ethnicity were considered along with the following covariates: maternal marital status at birth, parity, birthplace, age, education and tobacco use. Chi-square tests were used to identify associations between each covariate and the risk of preterm birth, as well as between each covariate and race and ethnicity (Arab American or non-Arab white). Logistic regression models assessed associations between Arab ancestry and risk of preterm birth.
Seventy-six percent of women were non-Arab white, 17% were black, 3% were Asian and 2% had Arab ancestry; 10% were foreign-born. Nearly three-quarters were married, 39% had had no children and 17% smoked tobacco. The prevalence of preterm births was 8% among Arab Americans and 9% among non-Arab white mothers, and the difference was statistically significant. In bivariate analysis, all covariates were associated with preterm birth; the proportion of such births was lower among foreig-nborn than among U.S.born mothers (8% vs. 10%), lower among married than among unmarried mothers (9% vs. 10%) and lower among mothers who did not smoke than among smokers (9% vs. 11%). The prevalence was 9–13% among the other subgroups, and it was elevated among women with three or more children, those older than 35 and those whose parity or education level was unknown.
All covariates were also associated with mothers' ethnicity: For example, Arab Americans had a lower prevalence of out-of-wedlock pregnancies than did non-Arab whites (4% vs. 28%), were less likely to smoke (3% vs. 16%) and were more likely to be foreign-born (81% vs. 7%).
The first set of regression analyses found that Arab Americans were less likely than non-Arab whites to have a preterm birth (odds ratio, 0.8), and this association remained when any one covariate was added as a control (0.7–0.9). In addition, the likelihood of having a preterm birth was decreased among foreign-born mothers (0.7), married mothers (0.8) and those who had had one child (0.9), and was elevated among those who smoked (1.2), those younger than 20 or older than 30 (1.1–1.5), those with a college education or less (1.04–1.2), those who had had two or more children (1.04–1.4) and those whose parity or education level was unknown (1.5–1.6). Additional calculations showed that ethnic differences in three covariates—maternal birthplace, marital status and tobacco use—explained most of the difference in risk of preterm birth between the two ethnic groups. In a model that included all of the covariates simultaneously, however, no association was found between Arab ancestry and risk of preterm birth.
The researchers note that their study was limited by the use of a narrow range of covariates, particularly for socioeconomic status, and reliance on clinical estimates of gestational age, which can be inaccurate for small infants. They also acknowledge that because the difference in the prevalence of preterm births between Arab Americans and non-Arab whites was small, the findings may have limited clinical significance, and that the findings may not be generalizable beyond Michigan. The researchers recommend that future analyses attempt to identify factors that may influence links between marital status and maternal birthplace and the risk of preterm birth, such as "differences in prenatal care habits, diet, attitudes toward childbirth, and levels of social cohesion between married and unmarried [mothers and between] American and foreign-born mothers."
—J. Thomas
REFERENCE
1. El-Sayed AM and Galea S, Explaining the low risk of preterm birth among Arab Americans in the United States: an analysis of 617,451 births, Pediatrics, 2009, 123(3):e438–e445.