Health and Behavior Among Latina Women and Their Children: Associations with Social Support, Perception of Safety, and Neighborhood Disorder
- Author(s): Lesar, Michelle Danit
- Advisor(s): Syme, Sherman L
- et al.
Health and Behavior Among Latina Women and Their Children:
Associations with Social Support, Perception of Safety, and Neighborhood Disorder
Michelle Danit Lesar
Doctor of Philosophy in Epidemiology
University of California, Berkeley
Professor (Emeritus) Sherman L. Syme, Chair
Social epidemiology focuses on the health effects of the social and physical environment and includes research on a broad range of topics, including social support and neighborhoods. The three chapters in this dissertation investigate social support, perception of neighborhood, and neighborhood physical disorder as they relate to maternal and child health. The study population consists of women and their children who are participants in the Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS) project, a study on the health effects of pesticide and allergen exposure in the Salinas Valley, an agricultural region in California. The data used in this dissertation come from primarily immigrant Latinos, an important yet understudied population.
In chapter one, I assessed the association between social support during pregnancy and the odds for cesarean delivery among Medi-Cal eligible Latinas living in the Salinas Valley. Although cesarean sections can be life-saving for women and infants in some situations, the proportion of births currently delivered via cesarean is unnecessarily high. The presence of a delivery companion is associated with lower cesarean frequency, and there is some evidence for an association between social support during pregnancy and reduced cesarean frequency. I analyzed data from 511 women who participated in the CHAMACOS project to test for an association between higher social support and lower odds for cesarean delivery. I measured social support via structural (marital status, number of close friends, church attendance, and living arrangement) and functional (Duke-UNC Functional Social Support Questionnaire and perceived partner support) types of support, and controlled for body mass index, socioeconomic position, acculturation, and alcohol, cigarette, or drug use. I used multiple logistic regression models with robust standard errors to assess the relationship between social support variables and the natural log odds of cesarean delivery. I did not find statistical evidence for an association between social support and the odds for cesarean delivery, as all 95 percent confidence intervals included the null value. However, some point estimates were large enough to suggest an association might exist, albeit in the opposite direction of what I hypothesized. Although my findings did not suggest an association between social support during pregnancy and cesarean delivery, I believe future observational and intervention studies are needed to shed further light on the relationship between social support and cesarean delivery among Latinas.
In chapter two, I explored maternal perception of neighborhood safety and its relationship to hours of television viewing among three-and-a-half-year-old Latino children from the CHAMACOS project. Due to its association with negative health outcomes (e.g., obesity), national guidelines support limiting children's time with entertainment media to no more than one to two hours per day. Parents who perceive their neighborhoods to be less safe may choose to keep their children entertained indoors with television, whereas parents who are less concerned about neighborhood safety may allow their children to play outdoors. I tested two hypotheses related to maternal perception of safety and children's television-viewing behavior: (1) children whose mothers perceive their current neighborhoods to be safer for outdoor play will watch less television than children of women who rate their current neighborhoods as less safe, and (2) there will be effect modification of the association in the first hypothesis by the child's sex, with the association being stronger for girls than boys. My study population consisted of 328 mother-child dyads. Television viewing was analyzed dichotomously based on national guidelines: two hours or less per day of television versus more than two hours per day. In addition to using overall daily television-viewing information, I also tested my hypotheses for weekend and weekday viewing. I assessed maternal perception of neighborhood safety for child outdoor play via a likert scale (ranging from "very safe" to "very unsafe"); I used the four-category variable in analyses when interaction was not present and a dichotomous version of the variable ("very safe" versus all other responses) when interaction was present. As in chapter one, I used multiple logistic regression models and robust standard errors to test my hypotheses. I controlled for the confounding effects of socioeconomic position; maternal body mass index, employment status, marital status, and depression; whether or not the child lives with one or more siblings; and time of year. I found that mothers who perceived their neighborhoods to be less safe were more likely to have sons who watched television more than two hours per day, both in general and on the weekends, than mothers who thought their neighborhoods were very safe. The proportion of daughters who watched television more than two hours per day on an average day or on a weekend day did not differ by maternal perception of neighborhood safety. Also, maternal perception of neighborhood safety was associated with weekday television viewing habits among children, but there was no trend and this association did not vary by sex.
Finally, in chapter three I assessed whether or not neighborhoods with more symbols of physical disorder would have a higher proportion of five-year-old children spending more than two hours per day with entertainment media. As in chapter two, I explored the possibility of effect modification by sex. I analyzed data from 311 Latino children from the CHAMACOS project using a dichotomous variable for screen time: children who spent more than two hours per day watching television or videos or playing video or computer games versus children whose screen time was two hours or less per day. I used three outcome variables: overall (a weighted average of weekday and weekend) screen time, weekend screen time, and weekday screen time. To measure physical symbols of neighborhood disorder, I included objective information (via block-face observations) on cars, garbage and litter, cigarettes and cigars, graffiti, odors, the condition of residential buildings, yards and gardens, and "for rent" and "for sale" signs. I used a principal components analysis on a polychoric correlation matrix to create one summary variable from twelve original variables that categorized neighborhoods as "low", "somewhat low", "somewhat high", and "very high" with respect to physical disorder. I also used multiple logistic regression models with robust standard errors to test my hypotheses. The covariates I included in my final models were the child's sex and age, family income and home ownership, maternal body mass index, parental emotional problem, whether or not the child lives with a sibling, parent-rated child's health, the time of year of data collection, acculturation, and whether the neighborhood is urban, suburban, or rural. In the overall and weekend screen time analyses, I did not find statistical support for effect modification by sex nor did I find support for main effects. However, I did find evidence of interaction in the weekday screen time analyses; there was a main effect for girls, but not boys. For girls, the main effect appeared to be due solely to the odds ratio comparing the "somewhat high" and "very high" neighborhood disorder groups and was in the opposite direction of what I expected. I hypothesized that higher neighborhood physical disorder would be associated with more screen time, but it appears that neighborhoods with "somewhat high" physical disorder had a higher proportion of girls who spent more than two hours per weekday with entertainment media than did neighborhoods with "very high" physical disorder. As my findings appear to suggest a possible negative association for girls with respect to weekdays, and no other significant findings, I believe future research is needed to fully explore the potentially complex relationship between physical symbols of neighborhood disorder and screen time among five-year-old Latino children.
This dissertation focused on two social epidemiologic research topics, social support and neighborhoods, using data from a primarily immigrant, Latina population. Although I did not find evidence to suggest that an association between social support during pregnancy and cesarean delivery exists, I believe there may be an association, perhaps only for certain subgroups of Latinas, and this needs to be further explored in future research endeavors. With respect to neighborhood, my findings from chapters two and three appear to be inconsistent, with the results from chapter two suggesting that mothers who perceived their neighborhoods to be less safe have three-and-a-half-year-old sons who watch more than two hours per day of television on average and on the weekends, and chapter three providing weak support for an association between higher neighborhood physical disorder and less weekday screen time among five-year-old girls. Thus, the nature of the association between neighborhood and screen time may be complex and vary according to the day of the week, gender, and age. Future research should be conducted to better understand these relationships, with attention paid to how children spend their indoor leisure time.