These studies have used birth weight rather than gestational age, which is a major limitation of this work. Differences in the rates of low birth weight according to race and ethnicity remain, with African-American mothers bearing a substantially higher risk than white mothers, even after individual and community-level factors are taken into account (Roberts, 1997). Some specific neighborhood-level characteristics that have been associated with birth weight and the risk of low birth weight include indicators of neighborhood economic deprivation and crime (Elo et al., 2001). Adverse neighborhood conditions have also been found to lessen the effects of protective factors, such as prenatal care (O’Campo et al., 1997).

Specifically, these studies should achieve the following:

• Develop strong theoretical models of the pathways from psychosocial factors, including stress, social support, and other resilience factors, to preterm delivery as a basis for ongoing observational research. These frameworks should include plausible biological mechanisms. Comprehensive studies should include psychosocial, behavioral, medical, and biological data.

• Incorporate understudied exposures, such as the characteristics of employment and work contexts, including work-related stress; the effects of domestic or personal violence during pregnancy; racism; and personal resources, such as optimism, mastery and control, and pregnancy intendedness. These studies should also investigate the potential interactions of these exposures with exposure to environmental toxicants.

• Emphasize culturally valid measures in studies of stress and preterm delivery to consider the unique forms of stress that individuals in different racial and ethnic groups experience. Measurement of stress should also include specific constructs such as anxiety.

• Expand the study of neighborhood-level effects on the risk of preterm birth by including novel data in multilevel models. Data that address this information should be made more available to researchers for such activities. Interagency agreements for the sharing of data should be reached to support the development of cartographic modeling of neighborhoods.