Neighbourhood contexts and low birthweight: social disconnection heightens single parents risks in Saskatoon.

Muhajarine, Nazeem ; Vu, Lan T.H.

Low birthweight (LBW) is a common and preventable public health concern that has been implicated in a range of negative infant and adult health outcomes. (1-5) Epidemiological and clinical studies have linked LBW with a variety of risk factors; (1,6-8) however, most studies have focused on individual-level factors, predominantly maternal characteristics, which explain only a small proportion of the overall variation in birthweight. In order to more fully understand the determinants of LBW, we need to look beyond maternal characteristics and behaviours to aspects of the physical, social, and economic environment that may impact LBW, either by shaping prenatal behaviour or through other pathways. (8) Elucidating the environmental or contextual factors that are related to LBW may also suggest new strategies for preventing this problem, consistent with health promotion's increasing focus on communitylevel interventions and policy approaches. (9-11)

Studies examining both individual and contextual factors have found that community characteristics (such as economic hardship or social support) exert an impact on LBW in addition to the customary maternal characteristics. (12-18) Important as these studies are, they generally fall short on three counts. First, research has tended to focus on socio-economic dimensions of communities, neglecting other characteristics such as physical environment, services, and social connectedness. Second, few studies have defined neighbourhoods or communities in ways that are meaningful to residents, relying instead on commonly available divisions, like census tracts or health regions. Third, few studies have examined the interaction between individual and contextual factors.

This study aims to understand how a range of neighbourhood characteristics influence LBW, independently and in interaction with individual factors, in the context of community-defined neighbourhoods.

METHODS

Source population and data extraction

The study employed a population sample of all singleton children born between April 1, 1992 and March 31, 1994 to mothers who lived and delivered in the city of Saskatoon, Canada. These 8,504 births were identified through the birth registry maintained by Saskatchewan's vital statistics branch. For each child, the neighbourhood in which the mother resided at the child's birth and whether she received income assistance were indicated. Data used to define neighbourhood characteristics were from Statistics Canada's 1991 Census as well as local sources such as the municipal planning department, and from purpose-built neighbourhood surveys. (19,20) The neighbourhoods themselves are geographic and social entities that are meaningful to Saskatoon residents. (21)

Measurement

Individual-level Variables

The dependent variable, LBW, was defined as less than 2500 grams. Independent variables at the individual level were: mother's marital status (married/common-law, single parent, and unknown), mother's age (less than 20 years, 20-40, and over 40), father's age (same groupings as mother), Aboriginal status of child, child's sex, family income status (defined as 'low income' if income support had been received from the provincial/federal governments during the child's birth year), mother's history of stillbirth, and parity (one live birth, two or more).

Neighbourhood Context

We used six variables, each measuring an underlying contextual domain at the neighbourhood level. Four of the six variable domains were derived from principal component analysis of multiple items relating to neighbourhood context (socio-economic disadvantage, physical condition, social disconnection, and availability and accessibility of programs and services for children), and two domains consisted of single items (population density and smoking prevalence). The underlying constructs indicated by each variable domain are among the key contextual factors in neighbourhoods that have been theorized previously to influence children's growth and development. (15,22,23) Appendix A presents the specific variable items comprising each neighbourhood domain. For example, social disconnection was measured by five items, with more disconnected neighbourhoods being those with lower levels of voter participation in recent civic and federal elections, greater transience, higher levels of ethnic diversity, and higher crime rates.

Statistical approach

Multilevel modeling for binary outcomes examined the independent effects of neighbourhood factors. Three hierarchical models were estimated using HLM software, each building on the previous model and thus increasing in complexity. (24) Step 1 included only the estimated neighbourhood-mean probability of LBW, which provides estimates of the variance in probability of LBW observed between and within neighbourhoods. Step 2 added all individual-level risk factors one at a time. If a significant variance component was observed, the variable was retained as a random effect; otherwise, the variable was constrained to be fixed across neighbourhoods. In step 3, six neighbourhood variables were added. The final model retains statistically significant individual- and neighbourhood-level variables.

RESULTS

The proportion of LBW babies was 4.7%, with 29% of the cohort born to single mothers, and 13% to families receiving income assistance. Other socio-demographic characteristics are given in Table 1.

Table 2 presents the steps in multilevel modeling and the variables included in the final model with the significant individual- and neighbourhood-level variance components associated with LBW. At the individual level, children born into households receiving income assistance, with a mother over 40, and whose mother had experienced one or more previous stillbirths were at greater risk for LBW. In contrast, children of mothers who had more than one previous live birth were less likely to be LBW.

At the neighbourhood level, two contextual variables were significantly associated with LBW over and above the effects of individual factors. Adjusted for all other individual- and neighbourhood-level variables, the likelihood of having a LBW baby among women living in the most socio-economically disadvantaged neighbourhoods was 1.34 times higher (95% CI 1.07, 1.68) than that among women living in relatively affluent neighbourhoods. Living in neighbourhoods with available and accessible child-related programs and services was associated with a lower risk of LBW (OR=0.98, 95% CI 0.96, 1.00).

Figure 1 maps the association between socio-economically disadvantaged neighbourhoods and the risk of LBW, adjusted for covariates in Table 2. Neighbourhood socio-economic disadvantage is shown in shaded polygons, each polygon representing a neighbourhood; the darker the shade, the higher the level of socioeconomic disadvantage. The shaded polygons are overlaid with circles, which indicate adjusted LBW rates; the larger the circle, the higher the adjusted rate of LBW. The map shows that neighbourhoods with higher levels of socio-economic disadvantage are generally associated with higher rates of LBW.

Results also indicated that the impact of single-parent status on LBW was attenuated by lower levels of neighbourhood social disconnection. As shown in Figure 2, in neighbourhoods with low levels of social disconnection (indicated, for example, by 10th percentile score), the risk of LBW among single mothers is lower (OR=0.89, 95% CI 0.72, 1.17) than in neighbourhoods with high levels of social disconnection (indicated by 90th percentile score) (OR=1.57, 95% CI 1.18, 1.93).

DISCUSSION

Using individual and contextual data in a defined population, we sought to understand the role of neighbourhood contexts in LBW. Controlling for individual risk factors, newborns in neighbourhoods with more available and accessible child-related programs and services were less likely to be LBW. In contrast, infants in socio-economically disadvantaged neighbourhoods were at increased risk for LBW. Most interestingly, the risk of LBW among infants born to single mothers was influenced by the level of social disconnection in their neighbourhood.

[FIGURE 1 OMITTED]

Neighbourhood socio-economic disadvantage in this study was defined as a high concentration of lower income families, education levels less than grade 9, unemployment, single parents, people of Aboriginal ancestry, and lower rates of home and car ownership. The independent association between neighbourhood socioeconomic disadvantage and LBW found in this study is consistent with results of other studies. (12-18) Previous studies have suggested that one pathway through which neighbourhood socio-economic status may influence residents' health status is the physical environment (e.g., housing quality, proximity to major thoroughfares, lack of parks). (23,25) In this study, however, we accounted for the physical condition of neighbourhoods; thus the association between neighbourhood socio-economic disadvantage and LBW was independent of the effect of neighbourhood physical environment. This result indicates that interventions directed at increasing economic opportunities in disadvantaged neighbourhoods may contribute to enhanced maternal and child health.

The measure of neighbourhood programs and services in this study reflects not just the existence but also the accessibility (e.g., transportation assistance, cost, operating hours) of the following types of programs and services: early education, parenting, parent relief, family support, birth/prenatal support, nutrition, child care, support for special needs children, and sports and recreational activities for children. Although programs enumerated in this study were allocated to a neighbourhood based on their physical location, many programs serve a population from a wider geographical area. However, not all residents have the means to access resources in other neighbourhoods (i.e., if they do not own a vehicle and public transportation is inadequate). Systematic evaluation is, of course, essential to assess the impact of specific programs and services on the rate of LBW and other birth outcomes.

Neighbourhood social disconnection moderated the association between single parent status and LBW. The measure of neighbourhood social disconnection in this study is conceptually related to constructs of social capital (26) and collective efficacy. (27) Previous studies have found neighbourhood support to be associated with infant birthweight,16 but to our knowledge this is the first study reporting moderating effects of neighbourhood social disconnection within a high-risk group such as single mothers. This finding adds to the growing evidence that social processes that enhance neighbourhood connectedness, trust and efficacy have a significant positive effect on health. Further studies are necessary to delve deeply not only into how social processes positively affect health outcomes, but also into how they may interact with individual-level factors to mitigate negative impact on health. (17,18)

Our measure of neighbourhood social disconnection, however, has some limitations. It was constructed for the purpose of this study from available indicators. While the selection of these indicators was based on research and theory, the measure would have been strengthened by a direct assessment of residents' sense of connection to their neighbourhood. Moreover, as a new measure, it would benefit from further testing of its validity.

Another potential limitation of the study is the time gap between the data collected for the measure of availability and accessibility to neighbourhood programs (2001) and the birth cohort itself. While it would have been ideal to have the birth cohort (1992-1994) and the assessment of programs match in time, it is unlikely that this discrepancy led to spurious results. Most neighbourhood programs that were enumerated were of the type that would have existed throughout the 1990s. This decade saw no major changes to the type of services directed at children and families in Saskatoon.28

One of the challenges of conducting this kind of multilevel research is deciding which individual-level factors to include in the analysis. (29) While it is important to control for potential confounders, there is also the risk of including factors that are part of the causal pathway through which neighbourhoods influence health outcomes, thereby leading to an underestimation of the role that neighbourhoods play. (30,31) In this study, the individual-level factors are ones which are generally related to LBW (mother's age, marital status, parity, previous birth outcomes), but are not likely to be influenced by the neighbourhood environment, unlike, for example, maternal smoking, poor diet or inadequate prenatal care. In studies involving multilevel variables, it is critical to carefully conceptualize mechanisms or causal pathways through which contextual factors affect individual outcomes, for these mechanisms need to inform the selection of variables, statistical modeling and interpretation of findings. (29-32)

In conclusion, we found neighbourhood contextual factors to be important determinants of LBW, over and above individual socioeconomic characteristics and mothers' previous birth outcomes. Two neighbourhood dimensions were independently related with a higher risk of LBW: socio-economic disadvantage and lower levels of program availability and accessibility. Furthermore, an interesting interaction was found between factors at the neighbourhood and individual level: single mothers living in socially connected neighbourhoods were less likely to have a LBW infant than those living in more disconnected areas. These findings suggest that interventions aimed at reducing LBW would be enhanced by targeting high-risk neighbourhoods as well as high-risk individuals in order to eventually reduce maternal and child health disparities.

Appendix A Summary of six neighbourhood characteristics included
in the analysis

Neighbourhood       Item Description                   Data Source
Domain

Socio-economic      Proportion of population with      Census 1991,
environment         Aboriginal ancestry                Statistics
alpha=0.94)         Proportion of families below       Canada
                    low-income cutoff
                    Proportion of population with
                    education level below grade 9
                    Proportion of population who
                    own their homes
                    Average number of cars
                    per person
                    Proportion of families
                    headed by single parent
                    Proportion of population
                    employed

Physical            Condition of neighbourhood,        Neighbourhood
condition           proportion of housing in need      observation
(Cronbach           of major repair, street width,     survey,
alpha=0.73)         road condition, appearance,        2001 (20)
                    noise level, number of
                    stoplights and crosswalks

Social              Proportion of voter                City of
disconnection       participation in most recent       Saskatoon (19)
(Cronbach           civic elections
alpha=0.84)         Proportion of voter
                    participation in most recent
                    federal elections
                    Proportion of population who       Census 1991,
                    moved in the past year             Statistics
                                                       Canada

                    Ethnic diversity index:            City of
                    Calculated by comparing the        Saskatoon (19)
                    neighbourhood population of a
                    given ethnic group to the
                    population of that ethnic group
                    across the whole city, and
                    summing up these proportions
                    across all ethnic groups. The
                    ethnic diversity index shows
                    the intensity of the presence
                    of ethnic groups in that
                    neighbourhood; the higher the
                    sum, the more diversified
                    the population

                    Crime incidence: Created from
                    Saskatoon City data; takes into
                    account reported property crimes
                    (break and entry), vandalism,
                    arson, etc.

Availability        Type of programs and services:     Neighbourhood
and accessibility   early education, parenting,        Programs and
of programs and     counseling, birth/prenatal,        Services
services            nutrition, child care, sport       Survey, (20)
                    and recreation                     Understanding
                                                       the Early Years,
                                                       Saskatoon, 2001

Population          Average number of persons per      Census 1991,
density             household in neighbourhoods        Statistics
                                                       Canada

Unhealthy           Proportion of current smokers      Tobacco Use
lifestyle "norm"    in neighbourhoods                  Survey,
                                                       Saskatoon
                                                       Health Region

Received: March 1, 2008

Accepted: September 25, 2008

REFERENCES

(1.) Institute of Medicine. Preventing Low Birth Weight. Washington, DC: National Academy Press, 1985.

(2.) McCormick MC. The contribution of low birth weight to infant mortality and childhood morbidity. N Engl J Med 1985;312(2):82-90.

(3.) McCormick MC, Brooks-Gunn J, Workman-Danniels K, Turner J, Peckham G. The health and development status of very low birth weight children at school age. JAMA 1992;267:2204-8.

(4.) Hack M, Flannery D, Schluchter M, Cartar L, Borawski E, Klein N. Outcomes in young adulthood for very low birth weight infants. N Engl J Med 2002;346(3):149-57.

(5.) Barker DJP. Fetal and Infant Origins of Adult Disease. London, UK: BMJ Publishing Group, 1992.

(6.) Kramer M. Determinants of low birth weight: Methodological assessment and meta-analysis. Bull World Health Organ 1987;65:663-737.

(7.) Eisner V, Brazie JV, Pratt MW, Hexter AC. The risk of low birthweight. Am J Public Health 1979;69(9):887-93.

(8.) Kramer MS, Seguin L, Lydon J, Goulet L. Socio-economic disparities in pregnancy outcome: Why do the poor fare so poorly? Paediatr Perinatal Epidemiol 2000;14:194-210.

(9.) Korenbrot CC, Moss NE. Preconception, prenatal, perinatal and postnatal influences on health. In: Smedley BD, Syme SL (Eds), Promoting Health: Intervention Strategies from Social and Behavioural Research. Washington, DC: National Academy Press, 2000;149-56.

(10.) Muhajarine N, Vu L, Labonte R. Social contexts and children's health outcomes: Researching across the boundaries. Critical Public Health 2006;16(3):205-18.

(11.) Galea S, Freudenberg N, Vlahova D. Cities and population health. Soc Sci Med 2005;60:1017-33.

(12.) O'Campo P, Xue X, Wang M, Caughy M. Neighbourhood risk factors for low birth weight in Baltimore: A multilevel analysis. Am J Public Health 1997;87:1113-18.

(13.) Roberts EM. Neighborhood social environments and the distribution of low birthweight in Chicago. Am J Public Health 1997;87(4):597-603.

(14.) Collins JW, Herman AA, David RJ. Very-low-birthweight infants and income incongruity among African American and white parents in Chicago. Am J Public Health 1997;87:414-17.

(15.) Pearl M, Braveman P, Abrams B. The relationship of neighborhood socioeconomic characteristics to birthweight among 5 ethnic groups in California. Am J Public Health 2001;91(11):1808-14.

(16.) 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):1-8.

(17.) Rauh V, Andrews H, Garfinkel R. The contribution of maternal age to racial disparities in birthweight: A multilevel perspective. Am J Public Health 2001;91(11):1815-24.

(18.) Jarveline M-J, Elliot P, Kleinschmidt I, Martuzzi M, Grundy C, Hartikainen AL, et al. Ecological and individual predictors of birthweight in a northern Finland birth cohort 1986. Paediatric Perinatal Epidemiol 1997;11:298-312.

(19.) City of Saskatoon. Neighbourhood Profiles, 6th ed. Saskatoon, SK: City of Saskatoon, 1998.

(20.) Muhajarine N, Delanoy S, Hartsook B, Hartsook L. Community Mapping for Children in Saskatoon. Saskatchewan: Saskatchewan Population Health and Evaluation Research Unit, 2003.

(21.) The Office of City Clerk, City of Saskatoon. City of Saskatoon Development Plan, Bylaw No. 7799. October 25, 2006.

(22.) Robert SA. Community-level socioeconomic status effects on adult health. J Health Soc Behav 1998;39(1):18-37.

(23.) Evans G, Kantrowitz E. Socioeconomic status and health: The potential role of environmental risk exposure. Annu Rev Public Health 2002;23:303-31.

(24.) Raudenbush S, Bryk A, Cheong YF, Congdon R. HLM5: Hierarchical Linear and Nonlinear Modeling. Lincolnwood, IL: Scientific Software International, 2000.

(25.) Troutt D. The Thin Red Line: How the Poor Still Pay More. San Francisco, CA: West Coast Regional Office, Consumers Union of U.S., 1993.

(26.) Kawachi I, Berkman LF. Social cohesion, social capital, and health. In: Berkman LF, Kawachi I (Eds.), Social Epidemiology. New York, NY: Oxford University Press, 2000;174-90.

(27.) Sampson R, Raudenbush SW, Earls F. Neighborhood and violent crime: A multilevel study of collective efficacy. Science 1997;227:918-23.

(28.) Muhajarine N, Vu L, Labonte R, Dodds L, Fell D, Kephart G. Community and Family Characteristics, Income Dynamics and Child Health Outcomes: Researching across the Boundaries. Technical Research Report. Saskatchewan: SPHERU, 2004.

(29.) Berkman LF, Glass T, Brissette I, Seeman TE. From social integration to health: Durkheim in the new millenium. Soc Sci Med 2000;51:843-57.

(30.) Pearce N. Traditional epidemiology, modern epidemiology, and public health. Am J Public Health 1996;86:678-83.

(31.) Diez-Roux AV. Multilevel analysis in public health research. Annu Rev Public Health 2000:21:171-92.

(32.) O'Campo P. Invited commentary: Advancing theory and methods for multilevel models of residential neighborhoods and health. Am J Epidemiol 2003;157:9-13.

Nazeem Muhajarine, PhD, [1,2] Lan T.H. Vu, MD, PhD [1]

Author Affiliations

[1.] Department of Community Health and Epidemiology, College of Medicine, University of Saskatchewan, Saskatoon, SK

[2.] Saskatchewan Population Health and Evaluation Research Unit, Saskatoon, SK *

* Current affiliation: Department of Epidemiology, Hanoi School of Public Health, Vietnam

Correspondence and reprint requests: Dr. Nazeem Muhajarine, Department of Community Health and Epidemiology and Saskatchewan Population Health and Evaluation Research Unit, Health Sciences Building, University of Saskatchewan, 107 Wiggins Road, Saskatoon, SK S7N 5E5, Tel: 306-966-7940, E-mail: [email protected]

Acknowledgements: The authors acknowledge the financial contribution of Canadian Population Health Initiative-CIHI, and thank colleagues in the Saskatchewan Ministry of Health, the Planning Department, City of Saskatoon, and the Saskatoon Health Region for sharing data. This study is based in part on non-identifiable data. The interpretations and conclusions contained herein do not necessarily represent those of the Government of Saskatchewan or Ministry of Health.

Table 1. Characteristics of the Birth Cohort (1992-94) and Prevalence
of Low Birthweight, Saskatoon (n=8504)

Variable                                                   Number (%)

Birthweight                 Low                              399 (4.7)
                            Normal                          8105 (95.3)
Mother's Marital Status     Married/Common-law              5831 (68.6)
                            Single parent                   2480 (29.2)
                            Unknown                          193 (2.3)
Father's Age (years)        20-40                           6951 (91.3)
                            <20                              251 (3.0)
                            >40                              413 (4.9)
Mother's Age (years)        20-40                           7597 (89.3)
                            <20                              835 (9.8)
                            >40                               72 (0.9)
Child's Aboriginal Status   Non-Aboriginal                  7543 (88.7)
                            Aboriginal                       961 (11.3)
Child's Sex                 Male                            4412 (51.9)
                            Female                          4092 (48.1)
Income Assistance Status    Low income                      1192 (13.1)
during Child's Birth Year   (received income assistance)
                            No income assistance            6604 (86.9)
History of Stillbirth       No prior stillbirth             8337 (98.0)
                            At least one prior stillbirth    167 (2.0)
Parity                      One live born baby              3541 (41.6)
                            Two or more live born babies    4963 (58.4)

Table 2. Estimated Coefficients and Odds Ratios (95% Confidence
Intervals) from Multilevel Modeling: Individual and Neighbourhood
Characteristics Regressed on Low Birthweight, Saskatoon (n=8504)

Variable                         Intercept Only Model Individual Model
(reference category in            ([beta]) (SE)      ([beta]) (SE)
italics)

Intercept                          -3.01 (0.06)      -3.09 (0.06)
Family income assistance                              0.57 (0.14) *
status (low income vs. no
income assistance)
Child's sex (female vs. male)                         0.08 (0.10)
Parity (1 live born                                   0.36 (0.15) *
vs. >1 live born)
Stillbirth (1 or more prior                           1.42 (0.21) *
stillbirth vs. no stillbirth)
Child's Aboriginal status                            -0.11 (0.27)
(Aboriginal vs. non-Aboriginal)
Mother's marital status (single                       0.39 (0.17) *
mother vs. married/common law)
Mother's age (<20 vs. 20-40 yrs)                     -0.12 (0.22)
Mother's age (>40 vs. 20-40)                          1.01 (0.40) *
Father's age (<20 vs. 20-40)                         -0.15 (0.35)
Father's age (>40 vs. 20-40)                          0.26 (0.19)
Neighbourhood socio-economic disadvantage
Neighbourhood programs and services availability and accessibility
Interaction between neighbourhood social disconnection and
   single mother status

Variable                               Neighbourhood Model
(reference category in            ([beta]) (SE)       Odds Ratio
italics)                                             (95% CI)

Intercept                          -3.13 (0.06)
Family income assistance            0.59 (0.15) *     1.80 (1.35, 2.42)
status (low income vs. no
income assistance)
Child's sex (female vs. male)
Parity (1 live born                 0.41 (0.14) *     1.51 (1.14, 2.00)
vs. >1 live born)
Stillbirth (1 or more prior         1.38 (0.19) *     3.99 (2.74, 5.80)
stillbirth vs. no stillbirth)
Child's Aboriginal status
(Aboriginal vs. non-Aboriginal)
Mother's marital status (single     0.16 (0.15)       1.17 (0.88, 1.57)
mother vs. married/common law)
Mother's age (<20 vs. 20-40 yrs)   -0.20 (0.20)       0.82 (0.55, 1.20)
Mother's age (>40 vs. 20-40)        1.20 (0.40) *     3.33 (1.55, 7.14)
Father's age (<20 vs. 20-40)
Father's age (>40 vs. 20-40)
Neighbourhood socio-economic        0.29 (0.11) *     1.34 (1.07, 1.68)
disadvantage                       [dagger]
Neighbourhood programs and         -0.02 (0.01) *     0.98 (0.96, 1.00)
services availability and         ([double dagger])
accessibility
Interaction between
neighbourhood social
disconnection and                  0.23 (0.10) *      1.25 (1.02, 1.53)
single mother status

* Significant variables (p<0.05)

([dagger]) Odds ratio correspond to '1-unit' change in the
independent variable in a continuous scale; when contrasts
are applied to the independent variable, the odds of low
birthweight for neighbourhoods with highest socio-economic
disadvantage (score at 90th percentile) is 1.34 compared to
neighbourhoods with lowest socio-economic disadvantage
(score at 10th percentile).

([double dagger]) Odds ratio for low birthweight for neighbourhoods
with the most programs and services for children (90th percentile)
is 0.98 compared to neighbourhoods with fewest programs and services
(10th percentile).

Figure 2. Odds ratio and 95% confidence bars, based on
multilevel final model in Table 2: level of
neighbourhood social disconnection heightens effects
of single mother on low birthweight

     Lower limit   Odds ratio   Upper limit
10th                0.89
20th                0.94
30th                1.02
40th                1.10
50th                1.12
60th                1.18
70th                1.29
80th                1.40
90th                1.57

Level of social disconnection (in percentile)

Note: Table made from line graph.