neighborhood effects on child injuries

Neighborhood effects on child injuries

3rd Conference of theInternational Society for Child Indicators

July 29, 2011University of York

Jim McDonellTracy Waters

Institute on Family and Neighborhood LifeClemson UniversityClemson, SC USA

This research was sponsored in part by a grant from The Duke Endowment

IntroductionChild injuries emerged as a relevant issue in the

field of prevention and public policy in 1980s

Advances in medicine led to fewer child deaths from disease (polio, measles, etc)

Shift in thinking and terminology◦ Accidents = random, caused by chance or fate,

unpreventable ◦ Unintentional injuries = explicable and

preventable

In the United States, unintentional injuries are the leading cause of child mortality and morbidity [Centers for Disease Control and Prevention (CDC), 2008]◦ 12,000 child fatalities annually◦ 9 million initial visits to the emergency

department◦ Fatal child injury rate: 15 per 100,000◦ Nonfatal child injury rate: 11,272 per 100,000

Introduction

IntroductionFalls are the leading cause of nonfatal childhood

injury in the United States (CDC, 2008) 2.8 million children injured Injury rate: 3,420 per 100,000 children

Falls account for a large proportion of child injuries throughout the world: United Kingdom, 40% (Haynes et al, 2003) New Zealand, 40% (Kypri et al, 2001) USA, 38% (CDC, 2007)

IntroductionTransportation related injuries are the leading

cause of unintentional child fatalities in the United States (CDC, 2008)◦ Transportation-related death rate = 9.8 per

100,000 Motor vehicle crash (occupant) = 4.6 per 100,000 Pedestrian death rate = 1.2 per 100,000

Recent decline in child pedestrian injuries (Doukas et al., 2010)

IntroductionChildhood injuries are related to a number of

population and environmental factors (Freisthler et al., 2008)◦ Number of female headed households◦ Adult to child ratio◦ Neighborhood disadvantage◦ Residential instability◦ Child care burden◦ Social capital

IntroductionAs many child injuries occur in or near the

home, the context of neighborhood has received increased attention◦ 24% of child injuries occurred on the street◦ 15% of child injuries occurred at a park,

playground, or sports facility (Haynes et al., 2003)◦ Schools and parks are the most common sites of

child injuries leading to litigation (Frost, 1995)

IntroductionDuring middle childhood (5 – 9 years of age),

children are at increased risk of falls, especially falls at the playground (Kypri et al., 2001)

Children have greater independent mobility starting between 7 – 9 years of age (Soori & Bhopal, 2001)

But how do neighborhood physical and social characteristics contribute to these child injuries?

IntroductionNeighborhood characteristics are also important

for understanding motor vehicle and child pedestrian injuries ◦ Number of parked cars on the street◦ Multi-family dwellings◦ Number of pedestrians observed (Agran et al., 1996)

On school days, 71% of child pedestrian injuries occur between 3 – 7 pm (Newbury et al., 2008)

IntroductionTraffic calming techniques, such as speed humps,

are effective in reducing child pedestrian injury (Tester et al., 2004)◦ Children living on a street with a speed hump

were significantly less likely to have a pedestrian injury

◦ Speed humps and other physical structures do not require policing and appear to be more effective than conventional deterrents

Again, more research is needed on the influence of neighborhood characteristics on child injuries

IntroductionThis study attempts to fill a gap in the

literature by exploring the relationship between both physical and social characteristics of neighborhoods and unintentional child injuries.

After an overview of the methodology, this presentation will highlight the resulting path models and conclude with implications for research, policy, and practice.

MethodsThe sample consisted of 244 neighborhoods in 132

census block groups. The neighborhoods were located in the Upstate and Midlands regions of South Carolina.

Convenience sample of neighborhoods

Neighborhoods were defined using GIS software. ◦ Aggregations of roads having an apparent geographic

relationship◦ Limited through road or arterial intersection◦ Bounded by natural or constructed features◦ Isolated from other road aggregations by distance

MethodsIllustration of sampled neighborhood

MethodsThree independent observations per

neighborhood◦ One weekday afternoon/evening observation ◦ One weekend day morning/early afternoon

observation◦ One “anytime” observation

Observations completed during warm weather months by driving and/or walking through neighborhood

MethodsNeighborhood Observation ScaleConstruct Factor # items Alpha

Physical appearance

Neighborhood physical appearance 7 .94School/park/public space physical appearance 5 .89

Social appearance

Neighborhood social appearance

5 .55

Indicated social engagement 4 .63Observed social engagement 3 .56Park/public space social engagement 3 .63

Safety Resident watchfulness 3 .66Neighborhood safety risk 4 .68

Initial results indicate acceptable reliability and validity (McDonell & Waters, 2010)

Items measured on 10 point Likert-type scaleExample:

Methods

Yards are poorly kept

Yards are well kept

1 2 3 4 5 6 7 8 9 10

Poorly kept = Lawn overgrown; property is dirty and unkempt; does not appear that attention is given to upkeep

Well kept = Clean; property apparently maintained; grass is cut; stairs/porch swept and clean.

MethodsChild injury rates were calculated using ICD-9 CM

coded hospital inpatient and emergency room discharge diagnoses.

Injury codes were provided by the South Carolina Office of Research and Statistics (ORS) at the census block group level.

Injury codes corresponded to the same time period in which neighborhood observations occurred.

Methods21 categories of injuries were collapsed into 9

categories:◦ Road vehicle injuries◦ Other vehicle injuries◦ Poisonings◦ Falls◦ Other accidents◦ Medical intervention◦ Suicide◦ Homicide◦ Other injuries

MethodsChild injury codes were calculated at rates

per 1,000 children

Rates were also calculated for children by gender and by age group

Analytic ApproachPath analysis models were created using AMOS

19.0

Measures of neighborhood physical and social characteristics were previously validated using confirmatory factor analysis. These 8 factors were treated as observed endogenous variables.

Child injury rates were also treated as observed endogenous variables.

Analytic ApproachGoodness of fit indices utilized:

◦ Non-significant chi square◦ Comparative Fit Index (CFI) > 0.9◦ Root Mean Square Error of Approximation

(RMSEA) < 0.05

Initial models included all 8 neighborhood constructs. Theory and modification indices guided adjustments to models.

Neighborhood type 69.3% residential only16.8% predominately residential 6.1% commercial only 5.7% predominately commercial 2.0% mixed

Housing type 53.3% single family detached16.4% duplex or row house 3.6% apartment/multiple occupancy 14.3%

mobile homes12.3% other

Neighborhood characteristics

People in 11.9% noneneighborhood 56.6% fewer than 5

25.0% 5 to 12 6.6% more than 12

Age distribution11.3% under 12 8.2% 13 to 1712.0% 18 to 2437.5% 25 to 4420.7% 45 to 64 8.7% 65 and older

Gender 37.7% female61.8% male

Neighborhood characteristics

Variable Mean SDRate of road vehicle injuries 7.49 3.50Rate of injuries due to falls 11.01 5.04Resident watchfulness .46 .20Neighborhood social appearance 1.04 .12Observed resident engagement 1.35 .15Condition of sidewalks 6.90 1.04Indicated resident engagement 4.22 1.18Neighborhood safety risk 1.00 .09Park/Public space social appearance 1.76 .22

Model variables

Path model for road vehicle injuries

χ2(13) = 20.48, p = .08CFI = .99RMSEA = .049

The model explains26% of the variancein child injuries fromroad vehicle accidents

.48

.16

-.29

.48

.18

.32

.25

.00Neighborhoodwatchfulness

e1

.39.75

e4Condition of sidewalks

.00

.56Neighborhood

social appearance e2

.26Road vehicle

injuriese8

.15Observed resident

engagement e3

.19Indicated resident

engagement e5

Park/public space social engagement e7

.00

.43

e6

.00

-.24

-.17

.02

-.22

-.12

-.10 Neighborhoodsafety


χ2(13) = 20.48, p = .08CFI = .99RMSEA = .049

.48

.16

-.29

.48

.18

.32

.25


e1

.39.75


.00

.56Neighborhood


.26Road vehicle

injuriese8


engagement e3


engagement e5


.00

.43Neighborhood

safety e6

.00

-.24

-.17

.02

-.22

-.12

-.10

Neighborhood social characteristics accounted for most of the explained variance

Road vehicle injuries arelower in neighborhoodshaving a better socialappearance and moreresident social engagement.


χ2(13) = 20.48, p = .08CFI = .99RMSEA = .049

.48

.16

-.29

.48

.18

.32

.25


e1

.39.75


.00

.56Neighborhood


.26Road vehicle

injuriese8


engagement e3


engagement e5


.00

.43Neighborhood

safety e6

.00

-.24

-.17

.02

-.22

-.12

-.10

However, observed resident engagement had a marginal direct effect in the opposite direction


χ2(13) = 20.48, p = .08CFI = .99RMSEA = .049

.48

.16

-.29

.48

.18

.32

.25


e1

.39.75


.00

.56Neighborhood


.26Road vehicle

injuriese8


engagement e3


engagement e5


.00

.43

e6

.00

-.24

-.17

.02

-.22

-.12

-.10 Neighborhoodsafety

The condition of sidewalks, a single item measure, was the only physical appearance factor having a significant effect

Road vehicle injuries were lower when sidewalks were in better condition


χ2(13) = 20.48, p = .08CFI = .99RMSEA = .049

.48

.16

-.29

.48

.18

.32

.25


e1

.39.75


.00

.56Neighborhood


.26Road vehicle

injuriese8


engagement e3


engagement e5


.00

.43Neighborhood

safety e6

.00

-.24

-.17

.02

-.22

-.12

-.10

Of the two safety measures, neighborhood watchfulness had an indirect effect while neighborhood safety risk had both a direct and an indirect effect

Neighborhoodsafety


χ2(13) = 20.48, p = .08CFI = .99RMSEA = .049

.48

.16

-.29

.48

.18

.32

.25


e1

.39.75


.00

.56Neighborhood


.26Road vehicle

injuriese8


engagement e3


engagement e5


.00

.43

e6

.00

-.24

-.17

.02

-.22

-.12

-.10

Road vehicle injuries are lower in neighborhoods with greater watchfulness and safety

The total effect ofwatchfulness was -.17 while the total effect of safety was -.21

-.17

-.21

Path model for falls χ2(5) = 4.66, p = .46CFI = 1.00RMSEA = .000

.37.73

Neighborhood social appearance e2

.53

Observed residentengagement e3

.14

Unintentionalfallse6

.34


.00

Park/public space physical appearance e4

.00


e1

-.28

.30

.01

-.20

.13

-.36 .33.43

-.38

The model explains34% of the variancein child injuries fromunintentional falls


.37.73


.53


.14


.34


.00


.00


e1

-.28

.30

.01

-.20

.13

-.36 .33.43

-.38

Interestingly, injuries due to unintentional falls increased when parks and public spaces had a more pleasing physical appearance

This likely indicates higheruse of parks and public spaces creating moreopportunities for injuries from falls


.37.73


.53


.14


.34


.00


.00


e1

-.28

.30

.01

-.20

.13

-.36 .33.43

-.38

Again, factors related to neighborhood social appearance account for most of the variance

Observed resident engagement has a small direct effect on injuries due to falls.


.37.73


.53


.14


.34


.00


.00


e1

-.28

.30

.01

-.20

.13

-.36 .33.43

-.38

Again, neighborhood watchfulness had an indirect effect on unintentional falls. The total effect of watchfulness was -.14

Injuries due to unintentionalfalls are lower in neighborhoods with higher levels of resident watchfulness

-.14

This research further demonstrates the importance of neighborhood context to children’s safety

This suggests that environmental modification is key to improving child safety

However, the typical approach to improving children’s safety is by modifying the physical neighborhood

This study shows that attending to neighborhood physical features alone is not sufficient to improve children’s safety

Conclusions

Neighborhood characteristics, social characteristics in particular, are significant indicators of the risk of injuries to children

In socially cohesive settings, caregivers are more likely to watch over neighbor children, perhaps taking action to protect children from harm

In addition, social activity increases surveillance opportunities; residents are more likely to notice dangers

Too, when residents know and spend time with each other, they are more likely to talk about potential threats to children’s safety.

Conclusions

Strategies to increase social exchange among neighbors are likely to go a long way to improving children’s safety.

Such strategies as family activity groups, resident buying clubs, communal meals, and the like are low cost and easy to implement

A neighborhood watch group is a good way to foster resident engagement while simultaneously increasing watchfulness.

Finally, more research is needed to better understand the effect of neighborhood social and physical characteristics across a broader range of child injuries.

Conclusions

neighborhood effects on child injuries

Documents

child indicatorsjuly

common sites of child

introductionchildhood

nonfatal child injury

child fatalities annually9

child pedestrian injury

fewer child deaths

pedestrian death rate