American Journal of Obstetrics and Gynecology (2005) 192, S22–9

www.ajog.org

Neighborhood context and reproductive health

Jennifer F. Culhane, PhD, MPH,a,* Irma T. Elo, PhD, MPAb

Department of Obstetrics and Gynecology, Drexel University College of Medicine, Philadelphia, Pa,a

and Department of Sociology, University of Pennsylvania, Philadelphia, Pab

KEY WORDSRace

Reproductive healthBirth outcomeNeighborhood context

Racial and social class differences in rates of preterm birth and other adverse outcomes areamong the most widely recognized and least well-understood phenomena in the study ofreproductive health. Individual-level characteristics have failed to account for such gradients.

Recently, researchers have begun to argue that health in general and reproductive outcomesspecifically are rooted in social inequalities. One area of such inequality may be residentialsegregation and the associated race/ethnic differences in exposure to adverse neighborhoodconditions. We review the empiric data that examine the association between neighborhood

conditions and reproductive health. We also review the major challenges that researchers facewhen trying to incorporate neighborhood-level variables into studies of health outcomes. Ourgoal is to stimulate further research that simultaneously considers social, economic, and biologic

determinants of reproductive health.� 2005 Elsevier Inc. All rights reserved.

Racial and social class differences in rates of pretermbirth (PTB) and other adverse outcomes are among themost widely recognized and least well-understood phe-nomena in the study of reproductive health. Black infantsare twice as likely as white infants to be born low birthweight and 3 times as likely to be very low birth weight.Similarly, blackwomen are twice as likely aswhitewomento deliver before term, and this gap has been remarkablyconstant over the past 40 years.1-3 These racial disparitiesin birth outcomes are also largely responsible for theO2-fold excess in infant mortality rates among black infants.4

It is also well known that lower class women, independentof their race/ethnicity, have about a two-fold greater risk

Supported by grants HD36462 from the National Institute of Child

Health and Human Development, and TS-626 from the Centers of

Disease Control and Prevention.

* Reprint requests: Jennifer F. Culhane, PhD, MPH, Department

of Obstetrics and Gynecology, Drexel University College of Medicine,

245 N 15th St, 17th Floor, Philadelphia, PA 19107.

E-mail: JFC92@drexel.edu

0002-9378/$ - see front matter � 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.ajog.2005.01.071

of pretermdelivery comparedwithwomen in higher socialclasses.5 Yet, despite disparities in socioeconomic statusbetween black and white women, socioeconomic statushas failed to account for the entire racial gap.6 It furtherappears that benefits of income and education are notconferred equally to black andwhite women. Infants whoare born to college-educated black women are at a muchgreater risk of PTB,7 low birth weight,7 and infant death,8

compared with infants born to college-educated whitewomen.

In addition, other individual-level attributes (such ashealth behaviors and psychosocial characteristics) havenot fully explained either the racial difference or the socialclass gradient in reproductive outcomes.6,9 In recentyears, the failure of individual-level characteristics toaccount fully for these differentials has focused renewedattention on the possible role of the broader socialcontext, which includes neighborhood conditions, inproducing adverse birth outcomes.10 We review potentialmechanisms through which neighborhood context mayinfluence reproductive outcomes, including plausible

Culhane and Elo S23

Figure Conceptual framework.

biologic mechanisms linking, neighborhood context toPTB. In addition, we discuss some of the theoretic andmethodologic challenges in carrying out these types ofstudies. Our goal is to stimulate further research inthis area and to draw attention to the importance oflinking broad social and economic contexts with biologicmechanisms in future studies of reproductive healthoutcomes.

Conceptual framework

Neighborhood context

It has been suggested that the identification of plausiblemechanisms through which neighborhood context influ-ences health outcomes is perhaps the most pressing issuein the advancement of our understanding of how resi-dential context translates into either good or bad health.11

In the Figure, drawing on previous studies, we outlinea conceptual framework that links neighborhood contextto adverse reproductive events highlighting importantintervening variables along this pathway.10-14 In thismodel, neighborhood conditions that are hypothesizedto influence health, either directly or indirectly, arefeatures of the neighborhood’s social environment,service environment, and physical characteristics.13

Social environment refers to the level of neighborhoodcohesion or disorganization, norms of reciprocity, civicparticipation, crime, socioeconomic composition, resi-dential stability, and related attributes. These character-istics are hypothesized to influence health outcomesthrough a number of potential pathways that include

availability of social support, adaptation of copingstrategies, and exposure to chronic stress.15-20 Previousstudies have found that women who live in violent,crime-ridden, and physically decayed neighborhoods aremore likely to experience pregnancy complications andadverse birth outcomes, after adjustment for a range ofindividual-level sociodemographic attributes and healthbehaviors.16,21 In addition, studies have further pointedto the potential importance of neighborhood-level socialrelations for reproductive health.15,16

Characteristics of community service environmentreflect the availability of goods and services, such asaccess to quality health care, grocery stores, recreationalfacilities, and police and fire protection. The availabilityof such services is likely to be affected by the degree ofpolitical organization that influences the residents’ abilityto demand public services and recruit private serviceproviders to their neighborhoods. Poor public andprivateservices may have a direct and indirect impact on anindividual’s health by making residents more susceptibleto intentional and unintentional injuries; by limitingaccess to quality health care, healthy foods, andrecreational opportunities; and by increasing crimerates.11-14,22,23 Finally, the quality of the physical envi-ronment, which includes exposure to toxicants, noise andair pollution, and quality of the housing stock and publicspace, could have direct effects on health.14,17 Theconcentration of adverse neighborhood conditions alongall 3 dimensions discussed above are often closely tiedto the clustering of socioeconomic disadvantage. Thus,not surprisingly, a number of studies have documenteda significant association between neighborhood-levelsocioeconomic disadvantage and birth outcomes.24-27

S24 Culhane and Elo

In addition, living in racially or ethnically segregatedneighborhoods may influence health over and aboveindividual-level attributes.16,28,29 Residential segregationhad led to race/ethnic differences in exposure to adverseneighborhood conditions, differences that are mostpronounced between white and black women. Blackwomen are more likely than white women to live inneighborhoods with poor municipal services, limitedaccess to quality health care, high rates of crime andviolence, and poor quality housing.30,31 Closely associ-ated with residential segregation are other forms ofracial discrimination with potential adverse health con-sequences, including adverse birth outcomes.32-36 Forexample, in previous studies, a positive associationbetween black/white segregation and infant mortalityrates has been documented for black infants.37,38 At thesame time, for other ethnic groups, living in ethnicenclaves may confer health advantages. For example,the case has been made that Mexican women, who livein predominantly Mexican immigrant neighborhoods,deliver babies with higher birth weight.16

Individual-level characteristics

Neighborhood characteristics may exert their influenceon reproductive outcomes indirectly by patterningindividual-level economic opportunities and health be-haviors. For example, neighborhood-level opportunitystructure may restrict or facilitate access to schooling,training programs, and employment opportunities andthus influence reproductive outcomes through a wom-an’s attained socioeconomic status.13,39 Thus, socioeco-nomic status disparities in birth outcomes may originatein part in neighborhood context that shape individuals’life chances.

Furthermore, social characteristics of neighborhoods,perhaps through shared cultural norms and values, maywell influence health behaviors associated with adversereproductive outcomes. For example, individual-levelsmoking patterns,40,41 alcohol consumption, and dietarypractices,42-46 which seem particularly relevant to thisdiscussion, have been significantly associated with area-level deprivation when controlling for individual attrib-utes. We suggest that, in addition to health behaviors,adverse conditions (such as high crime rates, housingabandonment, and even noise pollution) may act as eitheracute or chronic stressors that exert their influencethrough stress physiologic factors and are thus potentialintervening mechanisms between neighborhood contextand reproductive health. Geronimus,47 for example,has argued forcefully that long-term exposure to socio-economic disadvantage, which includes residence insocioeconomically disadvantaged neighborhoods, is det-rimental formaternal reproductive health and is one of thefactors that contribute to more adverse birth outcomesamong black women rather than white women.

Thus, we have also included exposure to acute andchronic stress as one of the hypothesized pathwaysthrough which neighborhood context may affect birthoutcomes. At the individual level, a growing body ofempiric evidence based on methodologically rigorousstudies of pregnant women of different ethnic, socioeco-nomic, and cultural backgrounds, supports the premisethat mothers who experience high levels of psychologic orsocial stress during pregnancy are at significantly in-creased risk for PTB (relative risk, 1.5-2.0), even after anadjustment is made for other biomedical, sociodemo-graphic, and behavioral risk factors.48,49 In addition, wefound that adverse neighborhood conditions, such ascrime, homelessness, and tax delinquency, were associ-ated significantly with risk of urogenital tract infection,which is one of the leading causes of PTB,50,51 even afteran adjustment is made for individual-level risk factors.52

Finally, we should note that neighborhood contextand individual characteristics may interact such thatindividual characteristics may exert greater influence incertain neighborhoods or that the effects of neighbor-hood context are more pronounced for subgroups ofwomen who are stratified by socioeconomic status, race/ethnicity, or other individual attributes. For example,a recent study in Chicago found that high perceivedlevels of neighborhood support were associated posi-tively with birth weight only for white infants and thata significant negative association between birth weightand neighborhood-level economic disadvantage wasdocumented for black infants.15 This association re-mained significant even after an adjustment for maternalcharacteristics and other neighborhood conditions.15

O’Campo et al11,24 found that the early initiation ofprenatal care did not have the same beneficial effectfor women who lived in disadvantaged neighborhoodsin Baltimore, Md, which raised the possibility thatprenatal care in deprived settings is unable to addressvarious risks that are associated with adverse birthoutcomes.

Biological mechanisms

As noted earlier, neighborhood conditions such as highrates of crime and abandoned and dilapidated housingare conceptualized as stressful exposures and thus arehypothesized to have similar physiologic consequencesas do more traditional individual-level experiences, suchas negative life events. Stress, both at the individual andneighborhood level, may affect PTB through 2 majorphysiologic pathways. The first is a direct neuroendo-crine pathway that ultimately results in premature and/or greater activation of the maternal-placental-fetalendocrine systems that promote parturition. The secondpotential mechanism is an immune/inflammatory path-way wherein maternal stress may modulate character-istics of systemic and local immunity to increase

Culhane and Elo S25

susceptibility to infection or the proinflammatoryresponse to an existing infection.52-54

The plausibility of the direct neuroendocrine and theneuroendocrine-immune interaction pathways suggeststhat stressful exposures may have physiologic con-sequences over and above their possible influence onhealth-related behaviors. As evidence accumulates thatindividual-level stressful exposures can ‘‘get under theskin,’’ it is not hard to imagine that dangerous andrundown neighborhoods may exert a similar effect. Itis therefore possible that neighborhoods can influencehealth outcomes throughdirect physiologic dysregulation.

Methodologic considerations

Although there is a growing body of research on theeffects of neighborhood conditions on reproductive andother health outcomes, the study of whether and howcommunity conditions influence health faces a numberof theoretic and methodologic challenges.

How should neighborhoods be defined?

Most studies that have examined the effects ofneighborhood conditions on health outcomes have usedadministrative or political boundaries to characterizeneighborhoods. In the United States, studies of repro-ductive and other health outcomes commonly havedefined neighborhoods with the use of US censustracts.17,24,55,56 Alternatively, some studies have usedcensus block groups19,25 or alternative neighborhoodboundaries.15,16 The choice of census-based administra-tive units has been driven largely by convenience and theavailability of decennial census data for a wide range ofpopulation and housing characteristics. However, theseneighborhood units may provide only a roughmeasure ofneighborhood context and therefore lead to incorrectspecification of the effects of neighborhood character-istics.12,57 The assignment of an erroneous value forexposure to a neighborhood condition because of anarbitrary geographic boundary is known as aggregationor zone effect, which stems from the fact that all residentsin a given neighborhood are assigned the same expo-sure.58-60 Census tracts, for example, were designedoriginally to encompass areas with a similar populationsize and similar socioeconomic, housing, and demo-graphic characteristics.61 There can be considerablevariation in characteristics within census tracts, however,and individuals who reside in the same tract mayexperience very different ‘‘neighborhood’’ conditions.The same may be true for the smaller level of censusaggregation known as block groups, which are subunitswithin census tracts, although these areas are likely to bemore homogenous in their characteristics. Nevertheless,recent study that examined the association betweencensus tract and block group level measures of socioeco-nomic status and rates of low birthweight, childhood lead

poisoning, andmortality rates concluded that census tractand block group level measures provided similar results.The same socioeconomic measures and health outcomesthat were based on zip code areas, however, were found tobe less effective.61-63 To overcome problems that areposed by the use of conventional census geography, 2recent studies have modeled neighborhood effects onbirth weight using boundaries that were delineated on thebasis of homogenous neighborhood clusters and knowl-edge of traditional Chicago neighborhoods,15,16 which isan approach that merits replication in other settings. Yet,surprisingly few studies have examined whether the wayin which the neighborhood is defined has a substantiveimpact on findings of neighborhood effects on healthoutcomes.

An additional challenge in the examination of theeffects of neighborhood conditions on reproductive andother health outcomes is the choice of an appropriatescale for the exposure of interest.58-60,64 The impositionof a fixed spatial scale (such as census block groups ortracts) for all types of neighborhood exposures may notbe appropriate. For example, if nearby conditions (suchas the concentration of abandoned building or brokenwindows65) is associated with health outcomes, hownear or far do the abandoned buildings or brokenwindows need to be? It also is likely that appropriateneighborhood boundaries will vary for different neigh-borhood conditions (such as availability of health careservices and retail stores vs exposure to violent crime).Thus, careful thought should be given to the spatialscale of neighborhood boundaries for alternative mea-sures of neighborhood context.11 Furthermore, higherlevels of aggregation (such as states, counties, and citiesor metropolitan areas) may be relevant and suggest theneed to consider multiple levels of community context.

In addition, the literature on neighborhood effectsemphasizes another related analytic issue, namely spa-tial autocorrelation, which has not been considered inmost empiric studies of neighborhood effects on birthoutcomes with the exception of Morenoff’s16 study ofbirth weight in Chicago. Spatial autocorrelation ispresent when nearby or adjacent neighborhoods havesimilar outcomes (often referred to as spatial depen-dency), such as PTB rates, that are not accounted forafter an adjustment is made for neighborhood charac-teristics (such as high crime rates or high rates ofneighborhood poverty).58,66,67 Although adjustment forneighborhood characteristics can reduce the impactof spatial autocorrelation by capturing some of thisspatial dependence, often significant unobserved spatialdependencies may remain that can generate correlationsamong outcome variables in adjacent neighborhoods.That spatial autocorrelation can be significant in theanalyses of reproductive outcomes in urban areas, evenwhen a host of individual-level and neighborhood-levelcharacteristics are included in the analysis is illustrated

S26 Culhane and Elo

by Morenoff.16 The results from this study show thatsocial environment beyond the woman’s own neighbor-hood (ie, in adjacent neighborhoods) exerted aninfluence on birth weight, suggesting that such interde-pendence should be considered in analyses of neighbor-hood effects at least in urban contexts.16 Anotherconsideration, although not directly related to whatdefinition of neighborhood should be considered, relatesto the potential bias in estimates of neighborhood effectswhen residential choice itself is not considered. Becauseindividuals are not assigned randomly to neighborhoodsand unobserved characteristics that influence the choiceof residence are also likely to influence health outcomes,the estimation of causal neighborhood effects is prob-lematic with observational data.68,69 The collection oflongitudinal data and the modeling of residential choicemay help shed light on some of these potential biasesand should be encouraged.70

How should neighborhood context be measured?

There are 2 types of data that can be used tocharacterize neighborhoods: (1) aggregate or derivedvariables that are created from individual-level charac-teristics of neighborhood residents and (2) structural orintegral variables that do not have equivalent measuresat the individual level (such as housing quality, theavailability of goods and services, air pollution, thepresence of toxic waste sites).11,57 Most studies haveused aggregate-level variables that were created fromindividual-level data that usually were represented asmeans, medians, and distributions of such character-istics as household income, poverty, public assistanceuse, educational attainment, unemployment rates, andrace/ethnic composition of neighborhoods.24-27,56 It isoften hypothesized that aggregates of individual-levelcharacteristics serve as proxies for structural variablessuch as service availability, housing quality, and levels ofcrime and violence.12 Relatively few studies have in-cluded the latter type of community-level measures suchas crime rates, housing quality, and the presence ofcommercial establishments.16,22,24,71 The relative ab-sence of structural or integral variables probably reflectsthe limited availability of these types of data and isa limitation in this field.

Researchers face several challenges in characterizingneighborhood context using readily available deriveddata. First, these data provide an incomplete depictionof actual neighborhood conditions. In other words, notall neighborhoods that are characterized as poor anddisadvantaged based on census variables are actuallyalike.16 Rates of crime and violence, the quality of socialnetworks, public and private services, and other relevantneighborhood characteristics may vary substantiallyacross what appear to be similarly disadvantaged areas.Data to develop measures of neighborhood context that

could differentiate between what appear to be homoge-neously socioeconomically disadvantaged neighbor-hoods are needed.11 For example, recent studies thathave incorporated measures of social support15,16 andviolent crime24,16 have demonstrated the usefulness of theinclusion of a more varied set of neighborhood measuresin the analyses of reproductive outcomes. Potentiallyuseful, but largely untapped, sources of neighborhood-level data are administrative records that are maintainedby local governments and institutions. In addition,community surveys that solicit information aboutresidents’ perceptions of neighborhood context are po-tentially useful data for understanding neighborhood-level dynamics.15,16 An additional source of informationis observational data that may be used to captureneighborhood characteristics that are otherwise unavail-able and that can provide ‘‘objective’’ measures ofneighborhood context and that are independent of theperceptions of neighborhood residents. An example ofthe these approaches is the Project on Human Develop-ment in Chicago neighborhoods, which solicited infor-mation on neighborhood quality from a survey of arearesidents and objective indicators of neighborhood-levelsocial and physical disorder.72

The second challenge facing researchers is an analyticproblem caused by the high correlation among numerousneighborhood-level variables (such as poverty, low levelsof education, poor housing quality). One possible solu-tion is to develop indices of related neighborhood-levelconditions,15,73 although such indices obscure the discreeteffect of each distinct component. In addition, the abilityto estimate the independent contribution of neighbor-hood context to health outcomes requires an adequateadjustment for individual-level characteristics. Withoutindividual-level data, contextual-level effects may reflectunmeasured individual-level differences rather than in-dependent effects of neighborhood context, althoughcareful thought should be given to which individual-levelfactors are likely to act as confounders and which factorsare considered intervening variables.39,57

Finally, to date, the emphasis in this field has been tostudy the association between a woman’s current neigh-borhood context and her reproductive outcomes. Possi-bly, reproductive disadvantage is established early in thelife course; therefore, awoman’s neighborhood exposuresin her early childhood or even during her own gestationmay be as important in shaping her reproductive out-comes as her current neighborhood exposures.64 Thisemphasis undoubtedly stems from the cross-sectionalnature of data that are used in most studies and thenotion that current neighborhood conditions are a fairlygoodproxy for past conditions.A recent study of past andpresent neighborhood conditions on mental health statuschallenges this assumption by empirically demonstratingthat a lagged effect of childhoodneighborhood conditionsis a better predictor of mental health status in early

Culhane and Elo S27

adulthood, compared with current neighborhood con-ditions.74 The need for a life course perspective has alsobeen recognized in the study of physical health out-comes.11,12,64

Comment

The notion that community-level conditions can pro-duce profound effects on host susceptibility to diseasederives in part from the long-standing existence ofstrong social class and race/ethnic gradients in healthand death and the inability of individual-level character-istics to account for such gradients.75 The emphasis oncommunity context has a long history in public health.Initially, the field of public health was dominated byconcerns with neighborhood variation in health out-comes and associations between community conditionsand infectious diseases. As chronic disease became themajor cause of morbidity and death, public health focusshifted from the community to the individual.12 It isonly relatively recently that neighborhood context hasagain begun to play an increasingly important role inepidemiologic theories of health.11,12,76 The question ofwhether neighborhood context influences health out-comes over and above individual characteristics has alsoreceived heightened attention in the sociologic literatureon health and death in recent years.13 Advances instatistical techniques that facilitate the modeling ofmultilevel influences and the growing interest in theuse of geographic information systems have also madeanalyses of community-level variation and influences onhealth outcomes more feasible.77-79

Clearly, additional work is required to investigateoptimal ways to conceptualize neighborhood and toexplore the extent to which the association betweenneighborhood conditions and health outcomes varydepending on how neighborhood boundaries are defined.Despite their convenience, political and administrativeunits may not be the most appropriate way to delineateneighborhood boundaries. Furthermore, it may well bethat different neighborhood exposures exert their in-fluence on health at varying levels of aggregation. Largercatchment areas, for example, may be appropriate for themeasurement of availability of goods and services (such ashealth care facilities, recreational opportunities andgrocery stores); smaller geographic units may be moreappropriate for the assessment of the quality of housingstock, crime, and social characteristics of neighborhoods.Thus, it may be important to incorporate multiple levelsof influence simultaneously, depending on the researchquestion and the theoretic framework that is used to guidethe analysis. Studies that pay explicit attention to theseissues and other methodologic considerations are neededto advance our understanding of the potential influencesneighborhood context exert on health outcomes.

References

1. Luke B, Williams C, Minogue J, Keith L. The changing pattern

of infant mortality in the US: the role of prenatal factors and

their obstetrical implications. Int J Gynaecol Obstet 1993;40:

199-212.

2. Singh GK, Yu SM. Infant mortality in the United States: trends,

differentials, and projections, 1950 through 2010. Am J Public

Health 1995;85:957-64.

3. Ventura SJ, Martin JA, Curtin SC, Mathews TJ, Park MM. Births:

final data for 1998. Natl Vital Stat Rep 2000;48:1-100.

4. Iyasu S, Becerra JE, Rowley DL, Hogue CJR. Impact of very low

birth weight on the black-white infant mortality gap. Am J Prev

Med 1992;8:271-7.

5. Parker JD, Schoendorf KC, Kiely JL. Associations between

measures of socioeconomic status and low birth weight, small for

gestational age, and premature delivery in the United States. Ann

Epidemiol 1994;4:271-8.

6. Hummer RA. Racial differences in infant mortality in the US: an

examination of social and health determinants. Soc Forces

1993;72:529-54.

7. McGrady GA, Sung JF, Rowley DL, Hogue CJR. Preterm delivery

and low birth weight among first-born infants of black and white

college graduates. Am J Epidemiol 1992;136:266-76.

8. Schoendorf KC, Hogue CJR, Kleinman JC, Rowley D. Mortality

among infants of black as compared with white college-educated

parents. N Engl J Med 1992;326:1522-6.

9. Hummer RA, Biegler M, De Turk PB, Forbes D, Frisbie WP,

Hong Y, et al. Race/ethnicity, nativity, and infant mortality in the

United Status. Soc Forces 1999;77:1083-118.

10. Residential Segregation and Health. In: Kawachi I, Berkman LF,

eds. Neighborhoods and health. Oxford, UK: Oxford University

Press; 2003. p. 265-88.

11. O’Campo P. Invited commentary: advancing theory and methods

for multilevel models of residential neighborhoods and health.

Am J Epidemiol 2003;157:9-13.

12. Diez-Roux A. Bringing context back into epidemiology: variables

and fallacies in multilevel analysis. Am J Public Health 1998;88:

216-22.

13. Robert SA. Socioeconomic position and health: the independent

contribution of community socioeconomic context. Annu Rev

Sociol 1999;25:489-516.

14. Macintyre S, Ellaway A, Cummins S. Place effects on health: How

can we conceptualize, operationally and measure them? Soc Sci

Med 2002;55:125-39.

15. Buka SL, Brennan RT, Rich-Edwards JW, Raudenbush SW, Earls

F. Neighborhood support and the birth weight of urban infants.

Am J Epidemiol 2003;157:1-8.

16. Morenoff JD. Neighborhood mechanisms and spatial dynamics of

birth weight. Am J Sociol 2003;108:976-1017.

17. Anderson RT, Sorlie P, Backlund E, Johnson N, Kaplan GA.

Mortality effects of community socioeconomic status. Epidemi-

ology 1996;8:42-7.

18. Cubbin C, LeClere F, Smith GS. Socioeconomic status and injury

mortality: individual and neighborhood determinants. J Epidemiol

Community Health 2000;54:517-52.

19. Diez-Roux AV, Nieto JF, Muntaner C, Tyroler HA, Comstrock

GW, Shahar E, et al. Neighborhood environments and coronary

heart disease: a multilevel analysis. Am J Epidemiol 1997;146:

48-63.

20. Taylor SE, Repetti RL. Health psychology: what is an unhealthy

environment and how does it get under the skin? Annu Rev

Psychol 1997;48:411-47.

21. Zapata BC, Robolledo A, Atalah E, Newman B, King MC. The

influence of social and political violence on the risk of pregnancy

complications. Am J Public Health 1992;82:685-90.

S28 Culhane and Elo

22. Yen I, Kaplan GA. Neighborhood social environment and risk of

death: multilevel evidence from the Alameda county study.

Am J Epidemiol 1999;149:898-907.

23. Yen IH, Syme SL. The social environment and health: a discussion

of the epidemiologic literature. Annu Rev. Public Health

1999;20:287-308.

24. O’Campo P, Xue X, Wang M-C, Caughy MO. Neighborhood risk

factors for low birth weight in Baltimore: a multilevel analysis.

Am J Public Health 1997;87:1113-8.

25. PearlM, Braveman P, Abrams B. The relationship of neighborhood

socioeconomic characteristics to birth weight among 5 ethnic

groups in California. Am J Public Health 2001;91:1808-14.

26. Rauh VA, Andrews HF, Garfinkel RS. The contribution of

maternal age to racial disparities in birth weight: a multilevel

perspective. Am J Public Health 2001;91:1815-24.

27. Roberts EM. Neighborhood social environments and the distribu-

tion of low birth weight in Chicago. Am J Public Health 1997;87:

597-603.

28. Williams DR, Collins C. Racial residential segregation: a

fundamental cause of racial disparities in health. Public Health

Rep 2001;116:404-16.

29. Acevedo-Garcia D, Lochner KA. Residential segregation and

health. In: Kawachi I, Berkman LF, editors. Neighborhoods and

health. Oxford, UK: Oxford University Press; 2003. p. 265-87.

30. Massey DS, Denton NA. American apartheid: segregation and

the making of the underclass. Boston (MA): Harvard University

Press; 1993.

31. Wilson WJ. The truly disadvantaged: the inner city, the underclass

and public policy. Chicago: University Chicago Press; 1987.

32. David RJ, Collins JW Jr. ‘‘Bad outcomes in black babies: Race or

racism?’’. Ethn Dis 1991;1:236-44.

33. Dressler WW. Culture, stress, and disease. In: Johnson TM,

Sargent CF, eds. Medical anthropology: contemporary theory

and method. New York: Praeger; 1990. p. 248-67.

34. Wise PH. Confronting racial disparities in infant mortality:

reconciling science and politics. Am J Prev Med 1993;31:7-16.

35. Krieger N, Rowley DL, Herman AA, Avery B, Phillips MT.

Racism, sexism and social class: implications for studies of health,

disease and well-being. Am J Prev Med 1993;9:82-122.

36. Geronimus AT. To mitigate, resist, or undo: addressing structural

influences on the health of urban populations. Am J Public Health

2000;90:867-72.

37. Laveist TA. Linking residential segregation to the infant-mortality

race disparity in US cities. Sociol Soc Res 1989;73:90-4.

38. Polednak AP. Trends in US urban black infant mortality, by

degree of residential segregation. Am J Public Health 1996;86:

723-6.

39. Diez Roux AV. The examination of neighborhood effects on

health: conceptual and methodological issues related to the

presence of multiple levels of organization. In: Kawachi I,

Berkman LF, editors. Neighborhoods and health. Oxford, UK:

Oxford University Press; 2003. p. 45-64.

40. Duncan C, Jones K, Moon G. Smoking and deprivation: are there

neighborhood effects. Soc Sci Med 1998;48:497-505.

41. Kleinschmidt I, Hills M, Elliott P. Smoking behavior can be

predicted by neighborhood deprivation measures. J Epidemiol

Community Health 1995;49(suppl):S72-7.

42. Diehr P, Koepsell T, Cheadle A, Psaty BM, Wagner E, Curry S.

Do communities differ in health behaviors? J Clin Epidemiol

1993;46:1141-9.

43. Diez Roux AV, Nieto FJ, Sorlie P, Caulfield L, Tyroler HA,

Watson RL, et al. Neighborhood differences in diet: the athero-

sclerosis risk in communities (ARIC) study. J Epidemiol

Community Health 1999;53:55-63.

44. Hart C, Ecob R, Davey Smith G. People, places and coronary

heart disease risk factors: a multilevel analysis of the Scottish heart

health study archive. Soc Sci Med 1997;45:893-902.

45. Ellaway A, Anderson A, Macintyre S. Does area residence affect

body size and shape? Int J Obes Relat Metab Disord 1997;21:

304-8.

46. Yen IH, Kaplan GA. Poverty area residence and changes in

physical activity level: evidence from the Alameda county study.

Am J Pub Health 1998;88:1709-12.

47. Geronimus AT. Black/white differences in the relationship of

maternal age to birth weight: a population-based test of the

weathering hypothesis. Soc Sci Med 1996;42:589-97.

48. Lobel M. Conceptualizations, measurement, and effects of pre-

natal maternal stress on birth outcomes. J Behav Med 1994;17:

225-72.

49. Hedegaard M, Henriksen TB, Sabroe S, Secher NJ. Psychological

distress in pregnancy and preterm delivery. BMJ 1993;307:

234-9.

50. Goldenberg RL, Culhane JF. Infection as a cause of preterm birth.

Clin Perinatol 2003;30:677-700.

51. Goldenberg RL, Hauth JC, Andrews WW. Intrauterine infection

and preterm delivery. N Engl J Med 2000;342:1500-7.

52. Culhane JF, Rauh VA, McCollum K, Elo I, Hogan V. Exposure

to chronic stress and ethnic differences in rates of bacterial

vaginosis among pregnant women. Am J Obstet Gynecol

2002;187:1272-6.

53. Wadhwa PD, Culhane JF, Rauh VA, Barve SS, Hogan VK,

Sandman CA, et al. Stress, infection and preterm birth: a biobeha-

vioral perspective. Paediatr Perinatal Epidemiol 2001;15(suppl):

17-29.

54. Wadhwa PD, Culhane JF, Rauh VA, Barve SS. Stress and

preterm birth: neuroendocrine, immune-inflammatory and vascu-

lar mechanisms. Matern Child Health J 2001;5:119-25.

55. Sastry N, Hussey JM. An investigation of racial and ethnic

disparities in birth weight in Chicago neighborhoods. Demography

2003;40:701-25.

56. LeClere FB, Rogers RG, Peters K. Neighborhood social context

and racial differences in women’s heart disease mortality. J Health

Soc Behav 1998;39:91-107.

57. Pickett KE, Pearl M. Multilevel analyses of neighborhood socio-

economic context and health outcomes: a critical review.

J Epidemiol Community Health 2001;55:111-22.

58. Cressie NAC. Statistics for spatial data. New York: Wiley; 1993.

59. Longley P, Batty M. Spatial analysis: modeling in a GIS environ-

ment. Cambridge, UK: GeoInformation International; 1996.

60. Wong D. Aggregation effects in geo-referenced data. In:

Arlinghous SL, editor. Practical handbook of spatial statistics.

Boca Raton (FL): CRC Press; 1996. p. 83-107.

61. Krieger N, Chen TJ, Waterman PD, Soobader M-J, Subramanian

SV, Carson R. Geocoding and monitoring of US socioeconomic

inequalities in mortality and cancer incidence: Does the choice of

area-based measure and geographic level matter? Am J Epidemiol

2002;156:471-82.

62. Krieger N, Chen JT, Waterman PD, Soobader M-J, Subramanian

SV, Carson R. Choosing area based socioeconomic measures to

monitor social inequalities in low birth weight and childhood lead

poisoning: the Public Health Disparities Geocoding Project (US).

J Epidemiol Community Health 2003;57:186-99.

63. Krieger N, Waterman PD, Chen JT, Soobader M, Subramanian

SV. Monitoring socioeconomic inequalities in sexually trans-

mitted infections, tuberculosis and violence: geocoding and choice

of area-based socioeconomic measuresdthe public health dispar-

ities geocoding project (US). Public Health Reports 2003;

118:240-60.

64. Macintyre S, Ellaway A. Neighborhoods and health: an overview.

In: Kawachi I, Berkman LF, editors. Neighborhoods and health.

Oxford, UK: Oxford University Press; 2003. p. 20-42.

65. Cohen D, Spear S, Scribner R, Kissinger P, Mason K, Wildgen J.

‘‘Broken windows’’ and the risk of gonorrhea [comment].

Am J Public Health 2000;90:230-6.

Culhane and Elo S29

66. Bailey TC, Gatrell AC. Interactive spatial data analysis. Essex,

UK: Longman Scientific & Technical; 1995.

67. Dubin RA. Spatial autocorrelation: a primer. J Housing Econ

1998;7:304-27.

68. Oakes JM. The (mis)estimation of neighborhood effects: causal

inference for a practicable social epidemiology. Soc Sci Med

2004;58:1929-52.

69. Diez Roux AV. Estimating neighborhood health effects: the

challenges of causal inference in a complex world [commentary].

Soc Sci Med 2004;58:1953-60.

70. Duncan GJ, Raudenbush SW. Assessing the effects of context in

studies of child and youth development. Educ Psychol 1999;34:

29-41.

71. Karvonen S, Rimpela A. Socio-regional context as a determinant

of adolescents’ health behaviour in Finland. Soc Sci Med 1996;

43:1467-74.

72. Raudenbush SW. The quantitative assessment of neighborhood

social environments. In: Kawachi I, Berkman LF, editors. Neigh-

borhoods and health. Oxford, UK: Oxford University Press; 2003.

p. 112-31.

73. Gordon G. Area-based deprivation measures: a UK per-

spective. In: Kawachi I, Berkman LF, editors. Neighborhoods

and health. Oxford, UK: Oxford University Press; 2003. p.

179-207.

74. Wheaton B, Clarke P. Space meets time: integrating temporal and

contextual influences on mental health in early adulthood. Am

Sociol Rev 2003;68:680-706.

75. Adler N, Matthews K. Health psychology: why do some

people get sick and some stay well? Ann Rev Psychol 1994;

45:229-59.

76. Von Korff M, Koepsell T, Curry S, Diehr P. Multi-level analysis

in epidemiologic research on health behaviors and outcomes.

Am J Epidemiol 1992;135:1077-82.

77. Bryk AS, Raudenbush SW. Hierarchical linear models: applica-

tions and data analysis methods. Newbury Park (CA): Sage

Publications; 1992.

78. DiPrete TA, Forristal JD. Multilevel models: methods and

substance. Ann Rev Sociol 1994;20:331-57.

79. Goldstein H. Multilevel statistical models. New York: Oxford

University Press; 1995.