Prenatal Health Behaviors andPsychosocial Risk Factors inPregnant Women of Mexican Origin:The Role of Acculturation

Ruth E. Zambrana, PhD, Susan C. M. Scrimshaw, PhD, Nancy Collins, PhD,and Christine Dunkel-Schetter; PhD

Introduction MethodsThe role of protective cultural behav-

iors as factors in the improvement of birthoutcomes is of interest in the UnitedStates as we have witnessed increasedrates of low birthweight.14 Attention hasfocused on Mexican-origin women who,despite economic disadvantage, have ratesof low birthweight similar to those ofnon-Hispanic Whites (5.7%).5-7 Impor-tant questions have emerged regarding thepathways through which acculturation, asa proxy of cultural norms, directly affectsrisk factors and indirectly affects birthoutcomes. Investigations show that Mexi-can immigrant women are more likelythan Mexican-American women to en-gage in prenatal health behaviors (absti-nence from alcohol, drug, and cigaretteuse during pregnancy)8-10 and to havepsychosocial assets (support from thebaby's father, availability of social net-works, and fewer stressful life events)associated with favorable birth out-comes.11-13 Less conclusive evidence sug-gests a link between positive maternalattitudes and favorable prenatal healthbehaviors.2'3 This study examined theeffects of these multiple variables within acommunity-based sample of Mexican-origin women. The specific researchquestions addressed were as follows:(1) Do Mexican immigrants and MexicanAmericans differ in prenatal health behav-iors and psychosocial behavioral riskfactors associated with adverse birthoutcomes? (2) Are measures of accultura-tion commonly used in public healthresearch associated with prenatal healthbehaviors and psychosocial risk factors?and (3) Do Mexican immigrants andMexican Americans differ in birth-weight and preterm delivery rates? Ifso, can these differences be accounted forby differences in the prenatal healthbehaviors and psychosocial risk factorsstudied?

Sample and Procedures

Two groups of women were re-cruited: Mexican Americans (40%; n =

366) and Mexican immigrants (60%; n =

545). Mexican Americans were born inMexico or the United States and hadresided in the United States since at least10 years of age. Mexican immigrantswere born in Mexico and had resided inthe United States for no more than 7 years.All respondents were primiparous, were17 to 35 years of age, had no more than 12years of education, and were at least at 20weeks' gestation at the time of theinterview (sample mean = 30 weeks).

Face-to-face interviews were con-ducted in 22 community-based prenatalcare clinics in Los Angeles County duringthe years 1987 through 1989. Potentialrespondents were approached by femaleinterviewers and screened for eligibility.Of those eligible, 96% agreed to partici-pate. Two thirds (64%) of the interviewswere conducted in Spanish. Followingdelivery, infant outcome data were re-trieved from medical records at 26 hospi-tals. Records were found and retrieved for78% of the sample. Our 22% loss rate isvery reasonable in survey research; withlow-income populations, however, it ispossible that the unmatched sample dif-fers systematically from the matchedsample. For example, women who did notdeliver at one of the study hospitals may

Ruth E. Zambrana is with the Social WorkProgram, George Mason University, Fairfax, Va.Susan C. M. Scrimshaw is with the School ofPublic Health, University of Illinois at Chicago.Nancy Collins is with the Department ofPsychology, State University of New York atBuffalo. Christine Dunkel-Schetter is with theDepartment of Psychology, University of Califor-nia at Los Angeles.

Requests for reprints should be sent to RuthE. Zambrana, PhD, George Mason University,Social Work Program, 4400 University Dr,Mailstop 2E8, Fairfax, VA 22030-4444.

This paper was accepted November 1,1996.

June 1997, Vol. 87, No. 6

Public Health Briefs

have had less stable family environments,more stress, higher medical risk, and moreadverse birth outcomes than the matchedsample. However, because there were nodifferences in match rates for MexicanAmericans and Mexican immigrants, anyacculturation differences observed in thisstudy are not likely to be due to systematicdifferences in match rates. This sampleincluded only those women who hadcomplete interview and medical recorddata.

Measures

Data were collected on demograph-ics, prenatal health behaviors, and psycho-social factors. Demographic variablesincluded age, income, marital status,insurance status, and living arrangements.Initiation of prenatal care was measuredby respondent report of number of weekspregnant at the first prenatal care visit.Substance use variables were measuredby respondent reports of smoking behav-ior (current smokers vs nonsmokers)alcohol use (current or prior alcohol usersvs nonusers), and illegal drug use (currentor prior drug users vs nonusers) 3 monthsbefore and during pregnancy. 14

Three stress measures were in-cluded. A 16-item life events inventoryassessed the number of life events that hadoccurred since the pregnancy.'5 The de-gree to which these events were distress-ing was measured by a single five-pointscale. An 8-item version of the PerceivedStress Scale assessed perceptions of strainduring the pregnancy'6 (alphas were .70for the English version and .75 for theSpanish version). The three stress mea-sures were standardized and averaged toform a single prenatal stress index formost analyses. A 6-item index measuringamount of support from the baby's fatherassessed positive and negative behaviorsduring the pregnancy (alphas were .90 forthe English version and .92 for theSpanish version).'7"18 Pregnancy attitudeswere included on an exploratory basis.2'3Respondents rated a series of items interms of their feelings about being preg-nant. Three items (feel special, feelhealthy, and feel lucky) were averaged toform an attitudes toward pregnancy index(ot = .50).

Acculturation was measured with the10-item version of the Cuellar scale.'9Items were standardized and averaged toform an acculturation index (ox = .87 inboth languages), with higher scores repre-senting more acculturation. As expected,Mexican Americans and Mexican immi-grants differed on this index (means of

TABLE 1-Selected Maternal Characteristics, Prenatal Psychosocialand Health Behaviors; and Infant Birth Outcomes forMexican-immigrant and Mexican-American Women

Mexican Immigrants(n = 545)

Maternal characteristicsMean age, yMean education, yMedi-Cal, %Live with baby's father, %Married, %Mean medical risk score

Prenatal psychosocial,and health behaviorsPrenatal stressSocial support from baby's fatherPositive attitudes toward pregnancyDrug use (ever), %Alcohol use (ever), %Current smoking, %Weeks to initiate prenatal care

Infant birth outcomesBirthweight, gGestational age (3-group)Gestational age, wk

21.818.18

14.0072.0051.001.24

-.1423.904.272.00

19.001.00

14.03

3363.672.90

39.69

Mexican Americans(n = 366)

20.08*10.49*49.00*52.00*35.00*1.55*

.05*22.05*4.06*7.00*

30.00*1.00

13.28

3341.102.90

39.91

*P< .001.

2.22 and 1.20, respectively; P < .001). Asa means of determining whether thisacculturation scale measured a singleunderlying construct or several relateddimensions, a factor analysis was con-ducted on the Cuellar items and twoadditional items: years of education andyears residing in Los Angeles. Two factorsemerged. The first, labeled integration inthe United States, included English lan-guage preference, literacy in English(reading and writing), years residingin Los Angeles, and education. Thesecond factor, labeled Mexican identity,included ethnic self-identification as Mexi-can, mother and father identification asMexican, and literacy in Spanish. Twoacculturation subscales, derived from theCuellar scale, were then formed by stan-dardizing and averaging the items on eachfactor (ts = .88 for integration and .70for identity).

Matemal medical risk and infantoutcome data were abstracted from medi-cal records via a standardized code book.A medical risk index was computed on thebasis of criteria from the Problem Ori-ented Perinatal Risk Assessment Sys-tem.20 The birth outcome variables werebirthweight (in grams) and gestational age(in completed weeks). Because of thesmall number of preterm deliveries, gesta-tional age was coded into three categories(clearly preterm [<35 weeks], marginally

preterm [36 or 37 weeks], or full term[>38 weeks]) for some analyses.

ResultsMexican immigrants and Mexican

Americans were compared on all studyvariables. As shown in Table 1, MexicanAmericans were younger, had completedmore education, were more likely to havepublic insurance (Medi-Cal), were lesslikely to be living with the baby's father,and were at higher medical risk thanMexican immigrants. There were nosignificant differences between the groupsin terms of annual income (samplemean = $11 058) or the percentage ofrespondents (20%) currently employed.

Mexican Americans reported moreprenatal stress, less support from thebaby's father, less positive attitudes to-ward their pregnancy, and more drug andalcohol use (see Table 1). There was nosignificant difference between the groupson initiation of prenatal care (samplemean = 13.73 weeks pregnant). As canbe seen in Table 1, there were no groupdifferences in gestational age or birth-weight. Among Mexican Americans, 7.7%of infants were delivered preterm ormarginally preterm, and 4.6% had lowbirthweights. Among Mexican immi-grants, 8.4% were preterm or marginally

American Journal of Public Health 1023June 1997, Vol. 87, No. 6

Public Health Briefs

TABLE 2-Correlations between Continuous Acculturation Indices andPrenatal Psychosocial Factors, Health Behaviors, and InfantBirth Outcomes: Total Sample

CuellarAcculturation Integration in Mexican

Index United States Identity

Prenatal stress .196** .172** -. 146**Positive attitudes toward pregnancy - .1 67** - . 54** .1 20**Social support from baby's father - . 25** - .141 ** .088*Weeks to initiate prenatal care -.106* -.119** .044Medical risk .1 03** .164* -.028Substance use .186** .143** -. 173**Drug use (ever) .238** .192** -.222**Alcohol use (ever) .1 91** .211** -.147**Current smoking .029 .011 -.025

Infant birth outcomesBirthweight -.010 -.012 .049Gestational age (3-group) - .013 -.017 .038

*P < .01; **P < .001.

TABLE 3-Intercorrelations among Prenatal Psychosocial Factors,Substance Use Index, and Medical Risk: Total Sample

1 2 3 4 5 6

1. Prenatal stress 1.0002. Social support from -.294*** 1.000

baby's father3. Positive attitudes toward -.303*** .1 92*** 1.000

pregnancy4. Weeks to initiate .000 -.069* .1 29*** 1.000

prenatal care5. Substance use index .202*** -.1 34** -.160*** -.024 1.0006. Medical risk .066* .004 -.016 -.047 -.032 1.000

*P<.05; **P< .1; ***P< .001

preterm, and only 2.6% had low birth-weights.

Correlational AnalysesWe examined links between our

three continuous measures of accultura-tion and all study variables (Table 2).Greater acculturation was associated withmore prenatal stress, less positive atti-tudes toward pregnancy, less social sup-port from the baby's father, earlier initia-tion of prenatal care, higher medical risk,and more drug and alcohol use. The threeacculturation indices showed the samepattem, with two exceptions. Increasedmedical risk and early initiation of prena-tal care were associated with more integra-tion in the United States, but not withMexican identity.

Next, we computed correlationsamong all of the prenatal variables (Table3). For this analysis, the three substance

use measures were standardized andaveraged to form a single index. Womenwho reported more prenatal stress had lesssupport from the baby's father, had lesspositive attitudes toward pregnancy, usedmore substances, and had higher medicalrisks. Women with more social supportfrom their baby's father had more positiveattitudes toward pregnancy, initiated pre-natal care earlier, and used fewer sub-stances. Finally, women with more posi-tive attitudes reported less substance useand initiated prenatal care later in preg-nancy.

Modeling Relationships betweenAcculturation, Psychosocial Factors,and Outcomes

In the last analysis, we performedstructural equation modeling to determinewhether there were indirect pathways (viathe prenatal factors) between accultura-

tion and birth outcomes. The hypoth-esized model included regression pathsfrom the two acculturation factors to eachof the prenatal variables and from each ofthe prenatal variables to the two birthoutcomes. Thus, links between accultura-tion and birth outcomes were expected tobe mediated by the prenatal variables. Theprenatal variables were allowed to freelycorrelate with each other, and a path fromgestational age to birthweight was alsoincluded because early delivery wasexpected to be a cause of low birthweight.

The model was analyzed via maxi-mum likelihood estimation.2' After estima-tion of the hypothesized model, the modelwas modified to improve fit. Nonsignifi-cant paths were dropped, and severalcorrelated residuals within the latentconstructs were added. These modifica-tions improved the fit but did not alter anyof the regression paths. The final model,presented in Figure 1, provided a good fitto the data according to a number ofindicators (comparative fit index = .958;chi square to degrees of freedom ra-tio = 2.3, df = 2; average standardizedresidual = .03). (For clarity, correlatedresiduals and correlations among prenatalvariables are not shown.)

Integration into the United Stateswas associated with more prenatal stress,which was associated with earlier deliv-ery. As expected, early delivery wasassociated with lower infant birthweight.Women who were more integrated alsohad less positive attitudes toward theirpregnancy, received less support from thebaby's father, had higher medical risks,and initiated prenatal care earlier. How-ever, these variables were not associatedwith birth outcomes. We also ran addi-tional models using the group variable(Mexican American, Mexican immigrant)instead of the acculturation factors, alongwith a substance use variable that in-cluded alcohol. Results were similar tothose presented here and did not changethe substantive conclusions.

DiscussionAlthough there is growing evidence

that psychosocial factors and health behav-iors during pregnancy are linked toadverse birth outcomes, we did not findconsistent relationships in the currentsample, with one exception: higher integra-tion in the United States was associatedwith higher prenatal stress, which wasassociated with preterm delivery. Thisprenatal stress, which was also associatedwith substance use and low social support,

1024 American Journal of Public Health June 1997, Vol. 87, No. 6

Public Health Briefs

reveals that low-income Mexican-Ameri-can women as a group are at risk forpreterm delivery. As Scribner has notedand we empirically tested, Mexican-American women, as they integrate intothe United States, experience a decreasein culture-specific protective factors thatare integrally related to the quality of thecommunity environments in which theylive.22 The pattern of differences in birthoutcomes by group is noteworthy. Therate of low birthweight for MexicanAmericans (4.6%) in this sample was

similar to the rate for Mexican-originwomen in Los Angeles County (4.9%),23and the low-birthweight rate among Mexi-can immigrants (2.6%) in this study was

almost one half that rate. Thus, combining

these groups in future studies may maskthe emerging group risk profile for Mexi-can-American women.

The processes that contribute to thedecrease in protective cultural behaviorsin Mexican-American women merit fo-cused investigation. Measures of accultura-tion, such as those used in this study andin public health research, assess a limitedset of socioeconomic characteristics (thecausal linkages between income, educa-tion, occupation, and language use) but donot measure the process of acculturationor change in cultural norms. Analyses ofacculturation measures show that Englishlanguage use is the dominant componentof acculturation.22 24 Future researchersneed to measure multiple factors-

including values, beliefs, and attitudestoward pregnancy and motherhood; prena-

tal health behaviors; and life stressors-toadvance knowledge of the role of culturaland community norms in protective andbehavioral risk factors for Mexican-originwomen. C]

AcknowledgmentsThis research was funded by the Agency forHealth Care Policy and Research (grant ROlHS/HD 05518-01 A) and by a National Instituteof Mental Health training grant (MH 15750)that supported Nancy Collins while collaborat-ing on the study.

We gratefully acknowledge the coopera-tion of the respondents, administrators, provid-ers, and medical record staff without whom thisstudy could not have been conducted.

American Journal of Public Health 1025

Number ofLife Events

.17

Note. Standardized coefficients are shown; all paths are significant at P - .05 (X2 = 347, df = 151, n = 91 1).

FIGURE 1 -Structural equation model of the relationships between two acculturation factors-prenatal psychosocialand health behaviors-and gestational age and birthweight.

June 1997, Vol. 87, No. 6

Public Health Briefs

References1. Troubling Trend: The Health ofAmerica's

Next Generation. Washington, DC: Na-tional Committee to Prevent Infant Mortal-ity; 1990.

2. Institute of Medicine. The Prevention ofLow Birthweight. Washington, DC: Na-tional Academy Press; 1985.

3. Institute of Medicine. Prenatal Care:Reaching Mothers, Reaching Infants. Wash-ington, DC: National Academy Press;1988.

4. Berendes HW, Kessel S, Yaffe S. Advancesin the Prevention of Low Birthweight.Washington, DC: National Center forEducation on Maternal and Child Health;1991.

5. Singh GH, Yu SM. Adverse pregnancyoutcomes: differences between US- andforeign-born women in major US racialand ethnic groups. Am J Public Health.1996;86:837-843.

6. Scribner R, Dwyer JH. Acculturation andlow birthweight among Latinos in theHispanic HANES. Am J Public Health.1989;79: 1263-1267.

7. Guendelman S, Gould JB, Hudes M,Eskenazi B. Generational differences inperinatal health among the Mexican Ameri-can population: findings from HHANES1982-84. Am J Public Health. 1990;80(suppl):61-65.

8. Moore J, Devitt M. The paradox ofdeviance in addicted Mexican Americanmothers. Genider Soc. 1989;3:53-70.

9. Marin G, Perez-Stables EJ, Marin B.Cigarette smoking among San FranciscoHispanics: the role of acculturation andgender. Am J Public Health. 1989;79:196-199.

10. Health United States 1995. Washington,DC: US Dept of Health and HumanServices; 1996. DHHS publication PHS96-1232.

11. Sherraden MS, Barrera R. Poverty, familysupport, and well-being of infants: Mexi-can immigrant women and childbearing.J Sociol Soc Wel. 1996;23(2):27-54.

12. Zambrana RE, Silva-Palacios V, Powell D.Parenting concerns, family support sys-tems, and life problems in Mexican-originwomen: a comparison by nativity. JCommunity Psychol. 1992;20:276-288.

13. Vega W. Hispanic families in the 1980s: adecade of research. J Marriage Fam.1990;52:1015-1024.

14. Belloc NB, Breslow L. Relationship ofphysical health status and mortality. PrevMed. 1972;2:67-8 1.

15. Golding JM. An Integrated Role Restric-tion and Stress Approach to GenderDifferences in Depression. Los Angeles,Calif: University of California at LosAngeles; 1985. Dissertation.

16. Cohen S, Kamarck T, Mermelstein R. Aglobal measure of perceived stress. JHealth Soc Behav. 1983;24:385-396.

17. Collins N, Dunkel-Schetter C, Lobel M,Scrimshaw S. Social support in pregnancy:psychosocial predictors of birth outcome

and postpartum depression. J Pers SocPsychol. 1993;65:1243-1258.

18. Dunkel-Schetter C, Sagrestano LM, Feld-man P, Killingsworth C. Social support andpregnancy: a comprehensive review focus-ing on ethnicity and culture. In: Pierce GR,Sarason BR, Sarason IG, eds. Handbook ofSocial Support and the Famil. New York,NY: Plenum Press; 1996; 1:375-412.

19. Cuellar I, Harris LC, Jasso R. An accultura-tion scale for Mexican American normaland clinical populations. Hispanic J BehavSci. 1980;2:199-217.

20. Hobel CJ, Youkeles L, Forsythe A. Prenataland intrapartum high-risk screening: riskfactors assessed. Am J Obstet Gynecol.1979; 135:1051-1056.

21. Bentler PM. EQS: Structural EquationsProgram Manual. Los Angeles, Calif:BMDP Statistical Software; 1989.

22. Scribner R. Editorial: Paradox as para-digm-the health outcomes of MexicanAmericans. Am J Public Health. 1996;86:303-304.

23. DeVane D, Gary AR, Elftman S, GallegosAA, Uretzky SA. Babies and Mothers atRisk: Perinatal Needs Assessment for LosAnigeles Counly Burbank, Calif: March ofDimes Birth Defects Foundations; 1990.

24. Cobas JA, Balcazar H, Benin MB, KeithVM, Chong Y. Acculturation and lowbirthweight infants among Latino women:a reanalysis of HHANES data with struc-tural equation models. Am J Public Health.1996;86:394-396.

1026 American Journal of Public Health June 1997, Vol. 87, No. 6