the association of urbanicity with infant sleep duration
TRANSCRIPT
Health & Place 18 (2012) 1000–1005
Contents lists available at SciVerse ScienceDirect
Health & Place
1353-82
http://d
n Corr
Health C
Boston,
E-m
journal homepage: www.elsevier.com/locate/healthplace
The association of urbanicity with infant sleep duration
Clement J. Bottino a, Sheryl L. Rifas-Shiman b, Ken P. Kleinman b, Emily Oken b, Susan Redline c,Diane Gold d, Joel Schwartz e, Steven J. Melly e, Petros Koutrakis e, Matthew W. Gillman b,f,Elsie M. Taveras a,b,n
a Division of General Pediatrics, Children’s Hospital Boston, MA 02115, United Statesb Obesity Prevention Program, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA 02115, United Statesc Division of Sleep Medicine, Brigham and Women’s Hospital, MA 02115, United Statesd Channing Laboratory, Department of Medicine, Brigham & Women’s Hospital and Harvard Medical School, MA 02115, United Statese Department of Environmental Health, Harvard School of Public Health, Boston, MA 02115, United Statesf Department of Nutrition, Harvard School of Public Health all in Boston, MA 02115, United States
a r t i c l e i n f o
Article history:
Received 26 January 2012
Received in revised form
18 May 2012
Accepted 9 June 2012Available online 28 June 2012
Keywords:
Sleep
Urbanicity
Population density
Infancy
Built environment
92/$ - see front matter & 2012 Elsevier Ltd. A
x.doi.org/10.1016/j.healthplace.2012.06.007
esponding author at: Department of Populati
are Institute and Harvard Medical School, 133
MA 02215, United States. Tel.: þ1 617 509 9
ail address: [email protected]
a b s t r a c t
Short sleep duration is associated with multiple adverse child outcomes. We examined associations of
the built environment with infant sleep duration among 1226 participants in a pre-birth cohort. From
residential addresses, we used a geographic information system to determine urbanicity, population
density, and closeness to major roadways. The main outcome was mother’s report of her infant’s
average daily sleep duration at 1 year of age. We ranked urbanicity and population density as quintiles,
categorized distance to major roads into 8 categories, and used linear regression adjusted for socio-
demographic characteristics, smoking during pregnancy, gestational age, fetal growth, and television
viewing at 1 year. In this sample, mean (SD) sleep duration at age 1 year was 12.8 (1.6) h/day. In
multivariable adjusted analyses, children living in the highest quintile of urbanicity slept �19.2 min/
day (95% CI:�37.0, �1.50) less than those living in the lowest quintile. Neither population density nor
closeness to major roadways was associated with infant sleep duration after multivariable adjustment.
Our findings suggest that living in more urban environments may be associated with reduced
infant sleep.
& 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Short sleep duration in infancy is prevalent (Armstrong et al.,1994) and associated with multiple adverse outcomes includingchildhood obesity (Bell and Zimmerman, 2010; Monasta et al.,2010; Taveras et al., 2008), behavioral and attention problems,(Zuckerman et al., 1987) maternal depression (Sadeh et al., 2010)and elevated parental stress (Meltzer and Mindell, 2007). Recentstudies suggest that the quality of sleep among young children isoften compromised, ($author1$ et al., National Sleep Foundation)and that pediatric sleep disturbances frequently become chronic,with few children outgrowing the problem (Kataria et al., 1987;Pollock, 1994). Several socio-environmental factors may influenceinfant sleep duration Nevarez et al., 2010). One such factor is thebuilt environment.
The built environment refers to aspects of a person’s surround-ings that are human made or modified including housing, urban
ll rights reserved.
on Medicine, Harvard Pilgrim
Brookline Avenue, 6th floor,
928; fax: þ1 617 509 9853.
(E.M. Taveras).
development, land use, transportation and commercial enter-prises (Papas et al., 2007). Built environment characteristics,assessed using geographic information systems (GIS), are asso-ciated with health outcomes and behaviors. For example, in astudy of over 400,000 births in Eastern Massachusetts, Zeka et al.found a risk of reduced birth weight associated with greaterexposure to traffic (Zeka et al., 2008). In a recent review of builtenvironment factors, mixed land use, housing development, andlevels of open space were all found to be associated with levels ofphysical activity among adults. (Durand et al., 2011). One advan-tage of GIS is the ability to categorize highly specific aspects of thebuilt environment. For example, Li et al. found that neighborhoodwalkability and fast food restaurant density in Portland, Oregoncorrelated with older residents’ blood pressure (Li et al., 2009). Inanother study of over 3000 subjects of the Framingham HeartStudy Offspring Cohort, closer proximity to fast food was asso-ciated with higher body mass index among women. (Block et al.,2011). To date, there have been few published studies exploringthe relationship between GIS variables that characterize the builtenvironment and child health outcomes, (Zeka et al., 2008; Sugliaet al., 2008) and none that we are aware of have examined sleepduration in infants.
C.J. Bottino et al. / Health & Place 18 (2012) 1000–1005 1001
The purpose of this study was to examine the associations ofcharacteristics of the built environment with sleep duration ininfancy. We were interested in examining environmental vari-ables related to greater noise, light, air pollution, and temperaturewhich can all adversely affect sleep duration. We hypothesizedthat increased urban land use, population density, and closerproximity to major roads would be associated with shorter sleepduration among infants at 1 year of age, and that socioeconomicstatus, biological and behavioral factors might confound thisrelationship.
2. Methods
2.1. Subjects/study design
The subjects for this study were participants in Project Viva, aprospective cohort study of gestational factors, pregnancy out-comes, and offspring health (Gillman et al., 2004). We recruitedwomen who were attending their initial prenatal visit from 1999to 2002 at 8 urban and suburban obstetrical offices of a HarvardVanguard Medical Associates, a multi-specialty group practice ineastern Massachusetts. Eligibility criteria included fluency inEnglish, gestational age less than 22 weeks at the initial prenatalclinical appointment, and singleton pregnancy. Details of recruit-ment and retention procedures are available elsewhere (Gillmanet al., 2004).
Of the 2128 women who delivered a live singleton infant, 2119provided their address at enrollment in the first trimester and1768 agreed to enroll their infants for follow up. We geocoded theresidential addresses to obtain our main exposures. Motherscompleted a mailed questionnaire at 1 year postpartum on whichthey reported child sleep duration. One-year sleep outcome datawere available for 1226 infants, who comprise the currentanalytic sample. Comparison of the 1226 participants in thisanalysis with the 902 excluded participants showed some differ-ences. For example, excluded vs. included participants were morelikely to be African–American (24.3% vs 11.0%), exposed tomaternal smoking during pregnancy (17.7% vs 9.6%) and live inhouseholds with annual income less than $40,000 per year (22.5%vs 11.2%).
Institutional review boards of participating institutionsapproved the study. All procedures were in accordance with theethical standards for human experimentation established by theDeclaration of Helsinki.
2.2. Measurements
2.2.1. Main exposure
From residential addresses collected at enrollment in the firsttrimester of pregnancy, we used a geographic information system(GIS) to determine our built environment exposures. All GISanalyses were performed using ArcGIS 9.3 (ESRI, Redlands, CA)in the Massachusetts State Plane projection, North AmericanDatum 1983. The variables included (1) urbanicity; (2) populationdensity; and (3) closeness to primary highways.
We defined urbanicity as the proportion of urban land usewithin 1 km of the mother’s residential address. This categoriza-tion makes use of nationwide land use data derived from satelliteimages with approximately 30 m resolution (the Multi-ResolutionLand Characteristics Consortium (www.mrlc.gov) 2001 NationalLand Cover Data Set). We considered low, medium and high-intensity developed land uses to be urban. Following an approachoriginally developed by Yanosky et al. for the Nurses’ HealthStudy (Yanosky et al., 2008), we used the ESRI ArcGISs SpatialAnalyst extension to calculate a surface with 30 m resolution
representing the proportion of urban land use within 1 km. Wethen interpolated values for urbanicity at each mother’s residen-tial address from this surface. Thus, urbanicity measures land useand defines urban vs. suburban environments and captures moreinformation than just simply the size of the population in acertain area.
We defined population density as the 2000 census block grouppopulation per square kilometer of dry land. To estimate distanceto major roadways, we used ESRI Streetmap 9.2 to calculate theminimum distance in meters from each geocoded address to thenearest major roadway. We defined major roadways based on USCensus Feature Class Codes and used in our analyses the roadsthat were potentially most associated with traffics, light, andnoise, e.g., A1 (primary roads with limited access) and A2(primary roads without limited access) roads.
2.3. Outcome measures
At 1-year postpartum, we asked mothers, ‘‘In the past month,on average, for how long does your child sleep in a usual 24-hperiod? Please include morning naps, afternoon naps, and night-time sleep.’’ Mothers provided responses in hours and minutes.Previous studies have recently validated parental report of infantsleep duration by associations of both actigraphic and parentalreport of sleep duration with elevated infant weight-for-length(Tikotzky et al., 2009).
2.4. Other measures
Using a combination of self-administered questionnaires andinterviews, we collected information about maternal age, educa-tion, parity, marital status, prenatal smoking (never, former,during pregnancy), household income, and child’s race/ethnicity.We obtained gestational age and infants’ birth weights frommedical records. We determined birth-weight-for-gestational-age (fetal growth) z-score according to a 1999–2000 US nationalbirth reference (Oken et al., 2003). At 1 year, we asked mothers toreport the number of hours and minutes their children watchedTV or videos in the past week.
2.5. Statistical analysis
We first examined descriptive statistics for our sample char-acteristics, exposures, and main outcomes. Next, we examined thebivariate relationships of our built environment exposures withour main outcome. We ranked urbanicity and population densityinto quintiles and categorized closeness to class 1 or 2 roads into8 categories to examine linearity of associations with sleepduration and because the distributions were highly right skewedfor all 3 exposures variables.
We used multivariable linear regression models to examinethe associations between each exposure variable in quintiles orcategories with sleep duration at 1 year. In multivariable models,we included only those covariates that were of a priori interest orconfounded associations of our main built environment exposureswith sleep duration. Model 1 was adjusted only for child sex. InModel 2, we additionally adjusted for socio-demographic char-acteristics, including maternal age, race/ethnicity, parity, educa-tion, marital status; and household income. In Model 3, wefurther adjusted for potential biologic and behavioral confoundersincluding maternal smoking during pregnancy; infant gestationalage and fetal growth, and television/video viewing at age 1 year.
Finally, because exposure to built environment variables mayvary by socioeconomic position, we investigated for effect mod-ification by maternal education and household income usingmultiplicative interaction terms, and we performed analyses
C.J. Bottino et al. / Health & Place 18 (2012) 1000–10051002
stratified by mothers’ education and household income. Weconducted all analyses using SAS, version 9.2 (Cary, NorthCarolina).
3. Results
At 1 year of age, children slept a mean of 12.8 (SD 1.6) h/day.Characteristics of urbanicity, population density, and closeness tomajor roadways are summarized in Table 1. Correlations among thebuilt environment exposures with each other and with infant sleepduration are provided in Table 1. Urbanicity and population densitywere strongly correlated (r¼0.73) but both were only weaklynegatively correlated with closeness to major roadways (Table 1).Other characteristics of the study participants are summarized inTable 2. Mothers living in census block groups with the highest v.lowest quintile of urbanicity were less likely to have a college degree(65% vs. 79%; chi-square p¼0.02), to have household incomes4$70,000 (56% vs. 81%; chi-square p¼o0.0001), to be married orco-habitating (91% vs. 98%; chi-square p¼0.003), and to be white(62% vs. 91%; chi-square p¼o0.0001). In addition, more whitemothers vs. mothers of non-white race/ethnicities lived in thehighest vs. lowest category of closeness to class 1 or 2 roads (84%vs. 62%; chi-square po0.0001).
Table 1Distributions of infant sleep duration at 1 year and measures of the built environmen
correlations. Data from 1226 participants in the Project Viva Cohort in Massachusetts.
Sleep duration at age 1 year(h/day)
Urbanicity (units)
Mean (SD) 12.8 (1.6) 779 (2 8 4)
Median (SE) 13 (0.05) 874 (8.1)
Range 5–19 0–1105
Built environmentExposures Sleep duration at age 1 year(h/day)
Urbanicity(quintiles)
Spearman correlation coefficients r (p-value)
Urbanicity �0.14 (o0.0001) 1.0
Population density �0.09 (0.002) 0.73 (o0.0001)
Closeness to class 1 or 2 roads 0.04 (0.14) �0.28 (o0.0001)
a Categories were o50 m, 50 to o100 m, 100 to o200 m, 200 to o400 m, 400 t
Table 2Characteristics of 1226 mothers and infants in Project Viva, according to quintile of ur
Quintile of urb
Overall 1
N (%) or mean (SD)
Urbanicity (unit) 779 (284) 326 (167)
Mother and family
Maternal age at enrollment (years) 32.3 (4.9) 33.3 (4.1)
Pre-pregnancy body mass index (kg/m2) 24.5 (5.1) 24.2 (4.4)
College graduate or more (%) 892 (73%) 213 (79%)
Household income 4$70,000/year (%) 768 (67%) 209 (81%)
Nulliparous (%) 613 (50%) 125 (46%)
Married or co-habitation (%) 1148 (94%) 266 (98%)
Mother smoked during index pregnancy (%) 114 (10%) 20 (8%)
White (%) 915 (75%) 246 (91%)
Child and home environment
Boy (%) 626 (51%) 141 (52%)
Birth weight (kg) 3.50 (0.57) 3.53 (0.58)
Gestational age at birth (weeks) 39.5 (1.8) 39.4 (1.9)
Birth weight for gestational age z-score (units) 0.22 (0.94) 0.32 (0.96)
Breastfeeding duration at 1 year (months) 6.4 (4.5) 5.9 (4.4)
Television viewing at 1 year of age (h/day) 1.2 (1.5) 1.3 (1.4)
Sleep duration at 1 year of age (h/day) 12.8 (1.6) 13.1 (1.6)
n P values are from chi-square for categorical variables and from ANOVA for contin
In bivariate analyses (Table 2), mean infant sleep duration waslower with higher quintile of urbanicity. Mean (SD) hours/day ofsleep was 13.1 (1.6) in quintile 1 and 12.5 (1.6) in quintile 5(ANOVA p¼0.001). We observed a similar overall pattern of lowersleep duration with higher quintile of population density. Mean(SD) hours/day of sleep was 13.0 (1.6) in quintile 1 and 12.7 (1.5)in quintile 5 of population density (ANOVA p¼0.03). In bivariateanalyses, children living closest to class 1 or 2 roads (o50 m) hadslightly shorter sleep duration than those living furthest away,but the associations of closeness to roads with infant sleepduration did not appear to be linear.
We show multivariable results in Tables 3 and 4. In Model 1,adjusted only for child sex, quintile of urbanicity was negativelyassociated with infant sleep duration. Being in quintile 5 vs.quintile 1 of urbanicity was associated with about over 30 min/day less sleep at 1 year (b �33.5 [95% CI �50.8, �16.2]).Adjustment for socioeconomic variables including maternal age,education, parity, marital status, and household income substan-tially attenuated the observed estimates (b �17.3 [95% CI �34.9,0.33] for quintile 5 vs. quintile 1). Further adjustment for smokingduring pregnancy, gestational age at birth, fetal growth, and TVviewing at 1 year of age did not materially affect the estimates(b �19.2 [95% CI �37.0, �1.50] for quintile 5 vs. quintile 1).Additionally, television viewing was an independent predictor of
t: (urbanicity, population density, and closeness to class 1 or 2 roads), and their
Population density (units) Closeness to class 1 or 2 roads (m)
4,597 (5,204) 1,730 (1,908)
2,703 (149) 1,078 (54.5)
67–42,100 10–12,941
Population density (quintiles) Closeness to class 1 or 2 roads (categoriesa)
1.0
�0.35 (o0.0001) 1.0
o o600 m, 600 to o1000 m, 1000 to o2000 m, Z2000 m.
banicity.
anicity
2 3 4 5 Pn
710 (68) 889 (37) 993 (24) 1072 (17)
32.4 (4.8) 32.3 (4.9) 32.4 (5.1) 31.2 (5.6) 0.0001
24.6 (5.3) 24.6 (4.9) 24.6 (5.3) 24.6 (5.4) 0.80
181 (73%) 183 (73%) 171 (73%) 138 (65%) 0.02
154 (66%) 148 (64%) 144 (67%) 109 (56%) o0.0001
106 (43%) 128 (51%) 119 (51%) 130 (61%) 0.002
236 (95%) 229 (92%) 215 (92%) 193 (91%) 0.003
23 (10%) 24 (10%) 19 (8%) 26 (13%) 0.39
199 (80%) 185 (74%) 147 (63%) 132 (62%) o0.0001
129 (52%) 122 (49%) 119 (51%) 111 (52%) 0.94
3.50 (0.56) 3.51 (0.51) 3.49 (0.61) 3.45 (0.58) 0.60
39.5 (1.8) 39.6 (1.7) 39.4 (2.0) 39.6 (1.8) 0.47
0.21 (0.96) 0.24 (0.89) 0.25 (0.94) 0.08 (0.96) 0.07
6.3 (4.5) 6.3 (4.4) 6.7 (4.7) 6.9 (4.6) 0.14
1.2 (1.5) 1.3 (1.6) 1.1 (1.4) 1.1 (1.3) 0.53
12.9 (1.5) 12.7 (1.6) 12.6 (1.7) 12.5 (1.6) 0.001
uous variables.
Table 3Associations of built environment exposures with infant sleep duration. Data From multivariable analysesa of 1226 infants participating in Project Viva.
Exposures N Estimates of sleep duration at 1 year (min/day), by multivariable model
Effect estimate (95% confidence interval)
Model 1 Model 2 Model 3
UrbanicityQuintile 1 271 0.00 (ref) 0.00 (ref) 0.00 (ref)
Quintile 2 248 �10.2 (�26.8, 6.44) �5.04 (�21.4, 11.35) �3.49 (�19.8,12.86)
Quintile 3 250 �20.0 (�36.6,�3.44) �11.2 (�27.7, 5.25) �11.4 (�27.8, 5.07)
Quintile 4 234 �29.2 (-46.1,�12.3) �15.4 (-32.3, 1.59) �11.9 (�28.9, 5.05)
Quintile 5 214 �33.5 (-50.8,�16.2) �17.3 (�34.9, 0.33) �19.2 (�37.0,�1.50)
Population densityQuintile 1 275 0.00 (ref) 0.00 (ref) 0.00 (ref)
Quintile 2 259 �19.3 (�35.7,�2.79) �14.6 (-30.7, 1.44) �13.1 (�29.2, 2.89)
Quintile 3 241 �22.1 (-38.8,�5.32) �11.5 (�28.0, 5.08) �9.17 (�25.7, 7.38)
Quintile 4 228 �25.7 (�42.8,�8.70) �8.00 (�25.3, 9.24) �8.25 (�25.5, 8.99)
Quintile 5 214 �17.9 (�35.2,�0.56) 1.69 (�15.9,19.29) �0.01 (�17.7,17.70)
Closeness to class 1 or 2 roadso50 m 45 0.0 (ref) 0.0 (ref) 0.0 (ref)
50 to o100 m 34 �9.18 (�52.4,33.98) �15.9 (�58.1,26.27) �24.4 (�66.3,17.37)
100 to o200 m 81 �16.4 (�51.7,18.89) �17.2 (�51.8,17.38) �19.8 (�54.3,14.70)
200 to o400 m 120 5.94 (�27.3,39.16) 3.82 (�28.8,36.39) �3.44 (�35.8,28.91)
400 to o600 m 122 �12.6 (�45.8,20.58) �15.8 (-48.4,16.72) �17.7 (�50.2,14.70)
600 to o1000 m 173 �23.2 (�55.0, 8.56) �25.9 (�56.9, 5.24) �28.1 (�58.9, 2.72)
1000 to o2000 m 274 �9.18 (�39.8,21.39) �12.2 (�42.2,17.73) �17.1 (�46.8,12.70)
Z2000 m 377 1.93 (�28.1,31.92) �6.65 (�36.2,22.87) �10.2 (�39.4,19.04)
a Model 1 is adjusted only for child sex. Model 2 is Model 1þrace/ethnicity, parity, education, household income, marital status, and maternal age. Model 3 is Model
2þsmoking during pregnancy, gestational age, fetal growth, and TV viewing at 1 year.
C.J. Bottino et al. / Health & Place 18 (2012) 1000–1005 1003
sleep duration: each 1 h/day increment in television viewing at1 year of age was associated with 5.89 min/day (95% CI: �9.66,�2.12) less sleep in a 24 h period. Offspring of black (b �43.1[95% CI �63.5, �22.8]), Hispanic (b �43.9 [95% CI �68.0,�19.8]) or other race/ethnicity mothers (b �21.4 [95% CI�41.1, �1.80]) slept fewer hours at 1 year of age than offspringof white mothers.
In models adjusted only for child sex, compared to children livingin the 1st quintile of population density, we observed shorter sleepduration in the 2nd through 5th quintiles of population density.However, after adjustment for socioeconomic variables, theobserved associations became null (Table 3). Closeness to class1 or 2 roads was not associated with infant sleep duration inunadjusted or multivariable adjusted models (Table 3).
We did not observe effect modification by maternal educationor household income for any of our exposure-outcome associa-tions (results not shown).
4. Discussion
In this prospective study of 1226 mother-infant pairs withinformation regarding the built environment in early life, wefound that infants who lived in more urbanized neighborhoodshad shorter sleep duration. The observed associations werepartially explained by adjustment for socio-demographic char-acteristics. Further adjustment for potential behavioral and bio-logical intermediates including fetal growth, gestational age,maternal smoking during pregnancy, and television viewing didnot substantially change the observed associations. Althoughliving in densely populated areas and living at shorter distancesto major roadways appeared to be associated with shorter sleepduration, these associations were largely explained by socio-economic characteristics of participants living in those areas.
To the best of our knowledge, this is the first study to report onassociations of aspects of the built environment with sleep
duration in infancy. While much is known about the role ofparental behavior and cognition in influencing infant sleep, less isknown about the social and environmental context influencingsleep in this age group (Sadeh et al., 2009; Tikotzky and Sadeh2009). Although the difference of 20 min in sleep duration mayappear small in magnitude, recent studies have shown that evensmall restrictions in infant sleep duration are associated withperceived difficult infant temperament, increased likelihood oflater behavior problems, compromised cognitive abilities andincreased body weight Sadeh et al., 2011). In addition, if lesssleep tracks throughout childhood, the observed small magni-tudes of effect could be additive throughout the child’s lifecourse.In this context of potential chronic sleep loss, even risk factorsduring infancy may be clinically relevant and merit awareness byclinicians. Our findings suggest that the built environment may beassociated with health even in the earliest stages of life.
Several potential mechanisms may underlie the associationbetween the built environment and sleep duration. Exposure toroad-traffic and transportation-related noise (Miedema and Vos,2007), electric lighting (Dumont and Beaulieu, 2007) and airpollution (Zanobetti et al., 2010) may reduce sleep duration.Noise, light, air pollution, and higher temperature are eachprominent features of the urban landscape (Babisch et al., 2001;Kyba et al., 2011; Patel et al., 2010) and may therefore representintermediate pathways in the relationship between urbanicityand sleep duration in infancy. Living in more urban settings mayalso produce stress through increased hypothalamic-pituitary-axis (HPA) activation, which may in turn exert downstreameffects on infant sleep architecture. In their analysis of stressand neighborhood environments in Philadelphia, Matthews andYang found a 50% increase in self-reported stress among residentsliving in a neighborhood with a hazardous waste processingfacility such as a landfill or incinerator (Matthews and Yang,2010). Prior work has also established road-traffic noise as acontributor to stress (Babisch et al., 2001), and children exposedto increased traffic-related noise have been found to have
Table 4Multivariable adjusted associations of urbanicity and all covariates in the final
model with infant sleep duration at 1 year (min/day).
Estimate (95% CI)
UrbanicityQuintile 1 0.0 (ref)
Quintile 2 �3.49 (-19.8, 12.86)
Quintile 3 �11.4 (�27.8, 5.07)
Quintile 4 �11.9 (�28.9, 5.05)
Quintile 5 �19.2 (-37.0, �1.50)
Mother and familyMaternal age at enrollment (years) �1.76 (�3.08,�0.44)
College graduate or more
No �2.59 (-16.8,11.59)
Yes 0.0 (ref)
Household income
o$40,000/year �16.8 (-37.2, 3.62)
$40,000–70,000/year �9.97 (-24.0, 4.07)
4$70,000/year 0.0 (ref)
Parity
0 �10.5 (�28.2, 7.33)
1 0.16 (�17.0,17.35)
2þ 0.0 (ref)
Married or co-habitation
No �23.6 (�49.0, 1.80)
Yes 0.0 (ref)
Maternal smoking status
Never 0.0 (ref)
Former 6.20 (�7.17,19.56)
Smoked during pregnancy 4.97 (�14.5,24.46)
Race/ethnicity
White 0.0 (ref)
Black �43.1 (�63.5,�22.8)
Hispanic �43.9 (�68.0,�19.8)
Other �21.4 (�41.1,�1.80)
Child and home environmentSex
Male �5.00 (�15.7, 5.71)
Female 0.0 (ref)
Gestational age at birth (weeks) 0.93 (�2.06, 3.92)
Birth weight for gestational age z-score (units) 1.54 (�4.46, 7.55)
Television viewing at 1 year of age (h/day) �5.89 (�9.66,�2.12)
C.J. Bottino et al. / Health & Place 18 (2012) 1000–10051004
elevated overnight cortisol (Evans et al., 2001) and higher restingsystolic blood pressure (Evans et al., 2001; Belojevic et al., 2008).Among preschool children, increased HPA activation is associatedwith elevated stress (Kryski et al., 2011) as well as poor qualitysleep (Hatzinger et al., 2008). In this study, however, we wereunable to measure noise, air pollution, temperature, ambient lightlevels, or HPA axis activation directly, so these mechanismsremain speculative.
We found that markers of socioeconomic position confoundedthe relationships of urbancity and population density with infantsleep duration. Socioeconomic position is highly interrelated withthe built environment (Boone-Heinonen et al., 2010) and bothindividual and neighborhood level socioeconomic position mayimpact sleep characteristics. At the individual level, low socio-economic position has been associated with higher levels ofchronic stress, with poor sleep quality helping to explain theobserved associations (Hawkley et al., 2011). Patel et al. havedescribed a ‘sleep disparity’, whereby poor sleep quality corre-lates strongly with poverty and racial/ethnic minority statusamong adults (Patel et al., 2010). From prospectively gathereddata on preschool children, Hale et al. reported later bedtimesamong Black and Hispanic children compared to white children,and found that low maternal education, greater household size,and poverty were associated with decreased use of bedtimeroutines (Hale et al., 2009). As with individual and householdlevel effects, socioeconomic position may adversely impact sleep
at the neighborhood level through neighborhood disadvantageand disorder. Residence in a neighborhood that is perceived asnoisy, unclean, and crime-ridden was associated with poorerquality sleep among a sample of Texas adults, (Hale et al., 2010)and severe neighborhood disadvantage was associated with asleep disorder, obstructive sleep apnea among a sample of 8 to 11year old children after adjustment for household level socio-economic position (Spilsbury et al., Sep 2006). Thus, socioeco-nomic position is a plausible and important confounder of theassociation between the built environment and sleep.
Our study had several strengths including our ability to adjustanalyses for a number of potential socioeconomic, biologic, andbehavioral variables that might be independently associated withsleep duration. Several limitations also exist. Although mothershad diverse racial/ethnic backgrounds, their education andincome levels were relatively high. Our results may thereforenot be generalizable to more socioeconomically disadvantagedpopulations. In addition, given the differences among the childrenincluded in the analysis and those for which we did not haveavailable outcomes data, our findings may be subject to selectionbias. Second, while we had measures of several built environmentexposures, we did not collect information on factors such as airpollution or noise levels that could help examine potentialmechanisms of action. Furthermore, we did not collect othermeasures of infant sleep quality, e.g., number of nighttimeawakenings, sleep location, bedtime routines, co-sleeping, crowd-ing, or parental perceptions of sleep, all of which may be relatedto sleep duration. Third, we measured sleep duration by parentalreport rather than using an objective measure of sleep such asaccelerometers or diaries. Finally, an additional limitation of anystudy that uses an observational design is that the associationbetween the primary exposure and the outcome of interest maybe mediated by factors not measured or not considered. Residualconfounding by variables we did not measure, e.g., noise, light, airpollution, stress, or maternal depression, could explain ourobserved associations.
In conclusion, the living in more urban environments isassociated with shorter infant sleep duration. Further studiesare needed to clarify mechanisms relating the built environmentto infant sleep characteristics. Considering the built environmentalong with family and individual behaviors may help informinterventions to support sufficient, good quality sleep duringinfancy.
Funding
This work was supported by grants from the National Institute ofMinority Health and Health Disparities (R01 MD003963), the Envir-onmental Protection Agency (RD83479801) and by the NationalInstitute of Environmental Health Sciences (P01 ES009825).
Acknowledgements
We thank the participants of Project Viva and our researchstaff for their assistance.
References
Armstrong, K.L., Quinn, R.A., Dadds, M.R., 1994. The sleep patterns of normalchildren. Medical Journal of Australia 161 (3), 202–206.
Bell, J.F., Zimmerman, F.J., 2010. Shortened nighttime sleep duration in early lifeand subsequent childhood obesity. Archives of Pediatrics & AdolescentMedicine 164 (9), 840–845.
Monasta, L., Batty, G.D., Cattaneo, A., et al., 2010. Early-life determinants ofoverweight and obesity: a review of systematic reviews. Obesity Reviews 11(10), 695–708.
C.J. Bottino et al. / Health & Place 18 (2012) 1000–1005 1005
Taveras, E.M., Rifas-Shiman, S.L., Oken, E., Gunderson, E.P., Gillman, M.W., 2008.Short sleep duration in infancy and risk of childhood overweight. Archives ofPediatrics & Adolescent Medicine 162 (4), 305–311.
Zuckerman, B., Stevenson, J., Bailey, V., 1987. Sleep problems in early childhood:continuities, predictive factors, and behavioral correlates. Pediatrics 80 (5),664–671.
Sadeh, A., Tikotzky, L., Scher, A., 2010. Parenting and infant sleep. Sleep MedicineReviews 14 (2), 89–96.
Meltzer, L.J., Mindell, J.A., 2007. Relationship between child sleep disturbances andmaternal sleep, mood, and parenting stress: a pilot study. Journal of FamilyPsychology 21 (1), 67–73.
National Sleep Foundation. Sleep in America. Princeton, NJ2008.Kataria, S., Swanson, M.S., Trevathan, G.E., 1987. Persistence of sleep disturbances
in preschool children. Journal of Pediatrics 110 (4), 642–646.Pollock, J.I., 1994. Night-waking at five years of age: predictors and prognosis. The
Journal of Child Psychology and Psychiatry 35 (4), 699–708.Nevarez, M.D., Rifas-Shiman, S.L., Kleinman, K.P., Gillman, M.W., Taveras, E.M.,
2010. Associations of early life risk factors with infant sleep duration.Academic Pediatrics 10 (3), 187–193.
Papas, M.A., Alberg, A.J., Ewing, R., Helzlsouer, K.J., Gary, T.L., Klassen, A.C., 2007.The built environment and obesity. Epidemiologic Reviews 29, 129–143.
Zeka, A., Melly, S.J., Schwartz, J., 2008. The effects of socioeconomic status andindices of physical environment on reduced birth weight and preterm births inEastern Massachusetts. Environmental Health 7, 60.
Durand, C.P., Andalib, M., Dunton, G.F., Wolch, J., Pentz, M.A., 2011. A systematicreview of built environment factors related to physical activity and obesityrisk: implications for smart growth urban planning. Obesity Reviews 12 (5),e173–182.
Li, F., Harmer, P., Cardinal, B.J., Vongjaturapat, N., 2009. Built environment andchanges in blood pressure in middle aged and older adults. PreventiveMedicine 48 (3), 237–241.
Block, J., Christakis, N., O’Malley, A., Subramanian, S., 2011. Proximity to foodestablishments and body mass index in the Framingham Heart Study Off-spring Cohort over 30 yearsAmerican Journal of Epidemiology.
Suglia, S.F., Wright, R.O., Schwartz, J., Wright, R.J., 2008. Association between lungfunction and cognition among children in a prospective birth cohort study.Psychosomatic Medicine 70 (3), 356–362.
Gillman, M.W., Rich-Edwards, J.W., Rifas-Shiman, S.L., Lieberman, E.S., Kleinman,K.P., Lipshultz, S.E., 2004. Maternal age and other predictors of newborn bloodpressure. Journal of Pediatrics 144, 240–245.
Yanosky, J.D., Paciorek, C.J., Schwartz, J., Laden, F., Puett, R., Suh, H.H., 2008. Spatio-temporal modeling of chronic PM10 exposure for the Nurses’ Health Study.Atmospheric Environment 42 (18), 4047–4062.
Tikotzky, L., de Marcas, G., Har-Toov, J., Dollberg, S., Bar-Haim, Y., Sadeh, A., 2009.Sleep and physical growth in infants during the first 6 months. Journal of SleepResearch.
Oken, E., Kleinman, K.P., Rich-Edwards, J.W., Gillman, M.W., 2003. A nearlycontinuous measure of birth weight for gestational age using a United Statesnational reference. BMC Pediatrics 3, 6.
Sadeh, A., Tikotzky, L., Scher, A., 2009. Parenting and infant sleep. Sleep MedicineReviews Jul 22.
Tikotzky, L., Sadeh, A., -2009. Maternal sleep-related cognitions and infant sleep:a longitudinal study from pregnancy through the 1st year. Child Development80 (3), 860–874.
Sadeh, A., Mindell, J.A., Owens, J., 2011. Why care about sleep of infants and theirparents? Sleep Medicine Reviews 15 (5), 335–337.
Miedema, H.M., Vos, H., 2007. Associations between self-reported sleep distur-bance and environmental noise based on reanalyses of pooled data from 24studies. Behavioral Sleep Medicine 5 (1), 1–20.
Dumont, M., Beaulieu, C., 2007. Light exposure in the natural environment:relevance to mood and sleep disorders. Sleep Medicine 8 (6), 557–565.
Zanobetti, A., Redline, S., Schwartz, J., et al., 2010. Associations of PM10 with sleepand sleep-disordered breathing in adults from seven U.S. urban areas. Amer-ican Journal of Respiratory and Critical Care Medicine 182 (6), 819–825.
Babisch, W., Fromme, H., Beyer, A., Ising, H., 2001. Increased catecholamine levelsin urine in subjects exposed to road traffic noise: the role of stress hormonesin noise research. Environment International 26 (7-8), 475–481.
Kyba, C.C., Ruhtz, T., Fischer, J., Holker, F., 2011. Cloud coverage acts as an amplifierfor ecological light pollution in urban ecosystems. PLoS One 6 (3), e17307.
Patel, M.M., Chillrud, S.N., Correa, J.C., et al., 2010. Traffic-related particulatematter and acute respiratory symptoms among New York City area adoles-cents. Environmental Health Perspectives 118 (9), 1338–1343.
Matthews, S.A., Yang, T.C., 2010. Exploring the role of the built and socialneighborhood environment in moderating stress and health. Annals ofBehavioral Medicine 39 (2), 170–183.
Evans, G.W., Lercher, P., Meis, M., Ising, H., Kofler, W.W., 2001. Community noiseexposure and stress in children. Journal of the Acoustical Society of America109 (3), 1023–1027.
Belojevic, G., Jakovljevic, B., Stojanov, V., Paunovic, K., Ilic, J., 2008. Urban road-traffic noise and blood pressure and heart rate in preschool children.Environment International 34 (2), 226–231.
Kryski, K.R., Smith, H.J., Sheikh, H.I., Singh, S.M., Hayden, E.P., 2011. Assessingstress reactivity indexed via salivary cortisol in preschool-aged children.Psychoneuroendocrinology 7.
Hatzinger, M., Brand, S., Perren, S., et al., 2008. Electroencephalographic sleepprofiles and hypothalamic-pituitary-adrenocortical (HPA)-activity in kinder-garten children: early indication of poor sleep quality associated withincreased cortisol secretion. Journal of Psychiatric Research 42 (7), 532–543.
Boone-Heinonen, J., Evenson, K.R., Song, Y., Gordon-Larsen, P., 2010. Built andsocioeconomic environments: patterning and associations with physicalactivity in U.S. adolescents. International Journal of Behavioral Nutrition andPhysical Activity 7, 45.
Hawkley L.C., Lavelle L.A., Berntson G.G., Cacioppo J.T. Mediators of the relation-ship between socioeconomic status and allostatic load in the Chicago Health,Aging, and Social Relations Study (CHASRS). Psychophysiology. 2011.
Patel, N.P., Grandner, M.A., Xie, D., Branas, C.C., Gooneratne, N., 2010. ’’Sleepdisparity’’ in the population: poor sleep quality is strongly associated withpoverty and ethnicity. BMC Public Health 10, 475.
Hale, L., Berger, L.M., LeBourgeois, M.K., Brooks-Gunn, J., 2009. Social and demo-graphic predictors of preschoolers’ bedtime routines. Journal of Developmen-tal & Behavioral Pediatrics 30 (5), 394–402.
Hale, L., Hill, T.D., Burdette, A.M., 2010. Does sleep quality mediate the associationbetween neighborhood disorder and self-rated physical health? PreventiveMedicine 51 (3-4), 275–278.
Spilsbury, J.C., Storfer-Isser, A., Kirchner, H.L., et al., Sep 2006. Neighborhooddisadvantage as a risk factor for pediatric obstructive sleep apnea. Journal ofPediatrics. 149 (3), 342–347.