Objective
To examine whether variation in neighborhood context is associated with preterm birth (PTB) outcomes and gestational age (GA) at delivery in Philadelphia, and to determine whether these associations might persist when considering relevant individual-level variables.
Study Design
We analyzed individual-level data collected for a prospective cohort study of singleton pregnancies with preterm labor. We merged block-group level data to each individual’s home address. Unadjusted analyses were performed to determine the association between block-group variables and individual-level outcomes. Block-group variables identified as potential risk factors were incorporated into multivariable individual-level models to determine significance.
Results
We analyzed data for 817 women. The prevalence of PTB <37 weeks was 41.5%. Although in unadjusted analyses several block-group variables were associated with PTB and GA at delivery, none retained significance in individual-level multivariable models.
Conclusion
Block-group level data were not associated with PTB outcomes or GA at delivery in Philadelphia.
Preterm birth (PTB) is a leading cause of perinatal morbidity and mortality. Well-recognized risk factors for PTB include obstetric history, maternal age, race, and socioeconomic status. However, the degree to which PTB risk might be affected by modifiable risk factors such as an individual’s environment, compared with nonmodifiable risk factors such as an individual’s genetics, remains unknown. Evidence from single-nucleotide polymorphism and genome-wide association studies have demonstrated significant but only nominally increased risk for PTB with specific allelic variations, suggesting a lack of evidence regarding a strong genetic causation in most cases of PTB. Other evidence suggests that an individual’s environment must play some role in PTB risk: PTB rates among African-American (18.4%) and European-American (11.7%) women are higher than PTB rates among African (11.9%) and European (6.2%) women, respectively.
The concept of possible “neighborhood effects” on health outcomes was introduced over 20 years ago. Since that time, there have been data from the nonobstetric literature to suggest an association between an individual’s geographic environment–specifically, one’s neighborhood–and chronic disease, mental health outcomes, cardiovascular mortality, serum cortisol levels, childhood obesity, and gynecologic disease. To date, however, there has been relatively little literature exploring the potential association between geographic environment–specifically, maternal neighborhood–and PTB. Although several studies have analyzed whether trends in neighborhood characteristics were associated with adverse perinatal outcomes including PTB, these studies were limited by their sole use of population-level data and provide somewhat conflicting results. Furthermore, they were unable to determine whether the relationship between geographic neighborhood and PTB might be modified by relevant individual-level variables.
To that end, our primary objective was to determine whether select geographic environmental variables related to maternal neighborhood are associated with prematurity. Specifically, we studied block-group level administrative (neighborhood crime, structural decline), census (population demographics, social stress, education, and employment), and survey (social capital) data to examine whether variation in individuals’ neighborhood context is associated with PTB outcomes and accounts for variation in gestational age (GA) at delivery in Philadelphia, PA. Our secondary objective was to determine whether any associations identified between environmental variables and PTB and/or GA at delivery might persist when considering relevant individual-level variables.
Materials and Methods
This study was a planned secondary analysis of data that were collected for a prospective cohort study at a single, urban tertiary care center between April 2009 and May 2012. The cohort consisted of women with periviable-preterm singleton pregnancies between 22- 33 6/7 weeks GA who presented to the labor and delivery triage unit with complaints concerning for spontaneous PTB, including contractions, cramping, vaginal bleeding, vaginal pressure, leaking vaginal fluid, and abdominal or back pain.
For the current study, we included all women from the parent study. Patients were excluded from the parent study and thus from these analyses for multiple-gestation, major fetal anomaly, intrauterine fetal demise, severe preeclampsia before enrollment, chronic steroid or immunosuppressive drug use, active immunologic disease, acute systemic febrile illness, and/or pregestational diabetes. Patients with unknown delivery information and/or whose home address was outside Philadelphia, PA, were also excluded from these analyses.
Patients were enrolled in the study by trained clinical research coordinators who obtained informed consent at the time of enrollment. The clinical research coordinators enrolled consecutive patients during daytime hours Sunday through Friday and during evening hours Monday through Thursday. Once a patient was enrolled in the study, all management decisions were made by the treating physician according to the standard of care at our institution.
Data collection
After enrollment, each patient was tracked for the remainder of her pregnancy and relevant delivery information was obtained through chart review. Pertinent demographic, medical, surgical, obstetric, gynecologic, and social histories were recorded. Maternal home address and zip code were also collected.
To examine whether variation in individuals’ neighborhood context is associated with PTB outcomes or accounts for variation in GA at delivery in Philadelphia, we collaborated with the University of Pennsylvania Cartographic Modeling Laboratory, a geographic information system spatial research center. The Cartographic Modeling Laboratory first used geographic information system industry standard software (ArcGIS 10.1; Redlands, CA) to convert, or geocode, each maternal address location into latitude and longitude coordinates that could be spatially joined with (merged) or into (aggregated) other geographic units. Maternal addresses were aggregated into census block-groups. Of note, census data are aggregated into block-groups, tracts, or other geographies based on the relative homogeneity of the population and economic characteristics. Block-groups are the smallest census geographic unit for which census population characteristics and economic status are tabulated. The average population within a block-group is approximately 1000 individuals.
Administrative (neighborhood crime, structural decline), census (population demographics, social stress, education and employment) and survey (social capital) data from 2003-2009 were collected from the Cartographic Modeling Laboratory’s databases. Administrative, census, and survey data along with maternal address data were aggregated into census block group boundaries using Summary Tape File Identification (STFID), a 12-digit unique identifier field for each specific census block group. All census block group-level address, administrative, census, and survey data were spatially joined, or merged, in ArcGIS using the STFID common field. The specific block-group variables that were analyzed are listed in Table 1 .
| Variable | P value (PTB) | P value (GA delivery) |
|---|---|---|
| Individual level variables | ||
| Prior PTB | < .001 | .002 |
| Prior LEEP/cold knife cone | .32 | .64 |
| African-American race | .42 | .12 |
| Maternal age | .003 | < .001 |
| Prenatal care | < .001 | .002 |
| GA first prenatal visit | .90 | .38 |
| Body mass index first prenatal visit | .91 | .82 |
| Gestational diabetes | .56 | .65 |
| Gestational hypertension | .02 | .06 |
| Chronic hypertension | .001 | .01 |
| Tobacco use | .009 | .002 |
| Cocaine use | .02 | .02 |
| Block-group level variables | ||
| Vacancy data, 2005 | ||
| Properties pending demolition | .61 | .07 |
| Properties clean and sealed | .79 | .60 |
| Lien sales for delinquent taxes | .66 | .87 |
| Vacant lots as listed by board of revision of taxes | .51 | .80 |
| Water service shutoffs | .93 | .34 |
| Vacancy data, 2006 | ||
| Properties demolished | .72 | .83 |
| Lien sales for delinquent taxes | .62 | .93 |
| Vacant lots as listed by board of revision of taxes | .57 | .95 |
| Water service shutoffs | .62 | .06 |
| Board of revision of taxes residential sales >$1000 | .52 | .19 |
| PHS greened lots | .41 | .82 |
| Total vacancy rate | .57 | .04 |
| Residential vacancy rate | .54 | .07 |
| Commercial vacancy rate | .13 | .19 |
| Vacancy data, 2007 | ||
| Properties demolished | .13 | .23 |
| Properties clean and sealed | .99 | .85 |
| Lien sales for delinquent taxes | .76 | .75 |
| Water service shutoffs | .78 | .39 |
| Vacancy data, 2009 | ||
| Vacant lots as listed by board of revision of taxes, 2009 | .45 | .83 |
| Board of revision of taxes residential sales >$1000 | .49 | .36 |
| Total vacancy rate | .28 | .01 |
| Residential vacancy rate | .46 | .04 |
| Commercial vacancy rate | .41 | .47 |
| Crime data, 2006 | ||
| Willful killing | .91 | .86 |
| Aggravated assaults | .56 | .87 |
| Burglaries | .49 | .60 |
| Burglaries and all thefts | .11 | .15 |
| Arsons | .63 | .98 |
| Other assaults | .11 | .04 |
| Transporting/receiving stolen goods | .41 | .66 |
| Vandalism and criminal mischief | .05 | .18 |
| Weapons violations of uniform firearms act | .53 | .53 |
| Prostitution | .57 | .31 |
| Sex offenses | .31 | .20 |
| All narcotics arrests | .44 | .75 |
| Disorderly conduct | .87 | .79 |
| Loitering and prowling | .64 | .68 |
| Illegal dumping | .86 | .85 |
| Crime data, 2009 | ||
| Willful killing | .87 | .63 |
| Aggravated assaults | .97 | .86 |
| Burglaries | .89 | .62 |
| Burglaries and all thefts | .40 | .12 |
| Arsons | .71 | .93 |
| Other assaults | .27 | .23 |
| Transporting/receiving stolen goods | .23 | .29 |
| Vandalism and criminal mischief | .24 | .73 |
| Weapons violations of uniform firearms act | .94 | .79 |
| Prostitution | .67 | .90 |
| Sex offenses | .33 | .80 |
| All narcotics arrests | .99 | .85 |
| Disorderly conduct | .37 | .32 |
| Loitering and prowling | .29 | .17 |
| Illegal dumping | .26 | .19 |
| Churches and block group party data | ||
| Places of worship within block group (2005) | .41 | .51 |
| Total block parties within block group (2003-2008) | .77 | .90 |
| Demographic and financial data, 2006 | ||
| Income disparity | .75 | .64 |
| Families in poverty | .37 | .14 |
| Female headed households with no husband and children 0-17 years old | .56 | .14 |
| Median household income, Asian | .99 | .46 |
| Median household income, black | .88 | .85 |
| Median household income, Hispanic | .91 | .39 |
| Median household income, native American | .50 | .64 |
| Median household income, Pacific | .90 | .98 |
| Median household income, biracial | .81 | .99 |
| Median household income, white | .72 | .42 |
| Median household income, white non-Hispanic | .84 | .56 |
| Median household income | .37 | .85 |
| Biracial household | .10 | .05 |
| Percent of population below 150% of poverty level | .59 | .67 |
| Population below 150% of poverty level | .25 | .05 |
| Average household size | .80 | .71 |
| Vacant housing units | .43 | .17 |
| Per capita income | .39 | .97 |
| Population density | .79 | .63 |
| White alone, total | .87 | .10 |
| Black alone, total | .98 | .96 |
| Asian alone, total | .30 | .07 |
| Hispanic alone, total | .03 | .02 |
| White alone non-Hispanic, total | .70 | .22 |
| Demographic and financial data, 2008 | ||
| Income disparity | .76 | .66 |
| Families in poverty | .34 | .13 |
| Female headed households with no husband and children 0-17 years old | .54 | .13 |
| Median household income, Asian | .99 | .46 |
| Median household income, black | .88 | .85 |
| Median household income, Hispanic | .91 | .39 |
| Median household income, native American | .50 | .64 |
| Median household income, Pacific | .90 | .98 |
| Median household income, biracial | .81 | .99 |
| Median household income, white | .72 | .42 |
| Median household income, white non-Hispanic | .84 | .56 |
| Median household income | .36 | .85 |
| Biracial household | .32 | .11 |
| Percent of population below 150% of poverty level | .57 | .66 |
| Population below 150% of poverty level | .22 | .05 |
| Average household size | .92 | .81 |
| Vacant housing units | .43 | .17 |
| Per capita income | .37 | .99 |
| Population density | .74 | .58 |
| White alone, total | .82 | .09 |
| Black alone, total | .94 | .99 |
| Asian alone, total | .30 | .06 |
| Hispanic alone, total | .03 | .02 |
| White alone non-Hispanic, total | .73 | .21 |
Data analysis
The primary outcome of our study was prematurity, assessed as both a dichotomous (PTB <37 weeks — present/absent) and a continuous (GA in delivery, weeks) variable. Pearson χ 2 analyses or Fisher exact as appropriate were used to determine associations between categorically measured individual-level risk factors and PTB, and nonparametric comparisons including Wilcoxon-Rank sum tests were performed to assess associations between the individual-level risk factors listed in Table 1 and both PTB and GA at delivery.
Factor analysis was also performed to determine whether the block-group variables could be combined to reduce the total number of covariates by describing factors (including neighborhood crime, structural decline, population demographics, social stress, education and employment, and social capital). Multiple variables from each category were considered in a factor analysis model. Principle factor method was applied, with the choice of the number of factors decided using the number of Eigen values >1. Factor results were rotated using an oblique rotation, and individual factors were estimated from each model. Estimates of each factor were then used as additional covariates and were examined for significant associations with PTB and GA at delivery, and are listed in Table 2 .
| Variables included in factor analyses | P value (PTB) | P value (GA delivery) |
|---|---|---|
| Vacancy data, 2005: properties pending demolition, properties clean and sealed, lien sales for delinquent taxes, vacant lots as listed by board of revision of taxes, water service shutoffs | ||
| Factor 1 | .76 | .49 |
| Vacancy data, 2006: properties demolished, lien sales for delinquent taxes, vacant lots as listed by board of revision of taxes, water service shutoffs, board of revision of taxes Residential sales >$1000, PHS greened lots, total vacancy rate, residential vacancy rate, commercial vacancy rate | ||
| Factor 1 | .62 | .94 |
| Factor 2 | .43 | .02 |
| Factor 3 | .75 | .19 |
| Vacancy data, 2007: properties demolished, properties clean and sealed, lien sales for delinquent taxes, water service shutoffs | ||
| Factor 1 | .86 | .86 |
| Factor 2 | .94 | .16 |
| Vacancy data, 2009: board of revision of taxes residential sales >$1000, total vacancy rate, residential vacancy rate, commercial vacancy rate | ||
| Factor 1 | .90 | .59 |
| Factor 2 | .50 | .04 |
| Crime data, 2006: willful killing, aggravated assaults, burglaries, burglaries and all thefts, arsons, other assaults, transporting/receiving stolen goods, vandalism and criminal mischief, weapons violations of UFA, prostitution, sex offenses, all narcotics arrests, disorderly conduct, loitering and prowling, illegal dumping | ||
| Factor 1 | .40 | .41 |
| Factor 2 | .21 | .51 |
| Factor 3 | .18 | .38 |
| Factor 4 | .44 | .60 |
| Factor 5 | .71 | .94 |
| Crime data, 2009: willful killing, aggravated assaults, burglaries, burglaries and all thefts, arsons, other assaults, transporting/receiving stolen goods, vandalism and criminal mischief, weapons violations of UFA, prostitution, sex offenses, all narcotics arrests, disorderly conduct, loitering and prowling, illegal dumping | ||
| Factor 1 | .81 | .57 |
| Factor 2 | .55 | .51 |
| Factor 3 | .99 | .97 |
| Factor 4 | .82 | .99 |
| Factor 5 | .79 | .95 |
| Churches and block group party data: places of worship within block group (2005), total block parties within block group (2003-2008) | ||
| Factor 1 | .91 | .62 |
| Demographic and income data, 2006: income disparity, families in poverty, female headed households with no husband and children 0-17 years old, median household income Asian, median household income black, median household income Hispanic, median household income native American, median household income Pacific, median household income biracial, median household income white, median household income white non-Hispanic, median household income, biracial household, percent of population below 150% of poverty level, population below 150% of poverty level, average household size, vacant housing units, per capita income, population density, white alone total, black alone total, Asian alone total, Hispanic alone total, white alone non-Hispanic total | ||
| Factor 1 | .98 | .28 |
| Factor 2 | .51 | .91 |
| Factor 3 | .42 | .01 |
| Factor 4 | .58 | .51 |
| Factor 5 | .19 | .51 |
| Factor 6 | .10 | .63 |
| Factor 7 | .82 | .54 |
| Factor 8 | .36 | .73 |
| Demographic and income data, 2008: income disparity, families in poverty, female headed households with no husband and children 0-17 years old, median household income Asian, median household income black, median household income Hispanic, median household income native American, median household income Pacific, median household income biracial, median household income white, median household income white non-Hispanic, median household income, biracial household, percent of population below 150% of poverty level, population below 150% of poverty level, average household size, vacant housing units, per capita income, population density, white alone total, black alone total, Asian alone total, Hispanic alone total, white alone non-Hispanic total | ||
| Factor 1 | .95 | .29 |
| Factor 2 | .44 | .81 |
| Factor 3 | .38 | .01 |
| Factor 4 | .58 | .46 |
| Factor 5 | .18 | .39 |
| Factor 6 | .30 | .98 |
| Factor 7 | .84 | .42 |
| Factor 8 | .34 | .82 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
