Relative strength of demographic, neighborhood, and school determinants in the prevalence of correct

1
Relative strength of demographic, neighborhood,
and school determinants in the prevalence of
correctable visual impairment The UCLA Mobile
Eye Clinic study

• Gergana Kodjebacheva
• February 11th, 2008
• CHS Doctoral Roundtable
• UCLA

2
Overview

• Prevalence and consequences of visual impairment
  (VI)
• Conceptual framework
• Research questions and hypotheses
• Methods
• Limitations and strengths
• Implications

3
Prevalence of Pediatric VI

• Definition compromised ability to resolve
  fineness in objects
• Most prevalent condition in childhood 20-25
• Common treatable/correctable conditions
• Refractive errors 15-20
• Amblyopia 1-5
• Racial/ethnic, age, gender, and income
  differences in prevalence rates

CDC 2002 Kemper 2004 Simons 1996 Zadnik 1997
Lennerstrand 1995
4
Negative consequences of VI

• Compromised cognitive, emotional, and physical
  development
• Irreversible vision loss
• Decreased interest in academic and recreational
  activities
• Failing schools poor reading scores
• Psychological individual and family stress
• Peer bullying
• Limited employment options in the future
• Personal and societal economic costs

5
Receipt of eye exams

• American Academy of Pediatrics recommendations
  call for regular eye exams as early as the
  infancy period.
• Still
• Only 30 under age 6 have ever had an eye exam.
• A tenth of children aged 6-17 years had an eye
  exam in the past year (CDC, 2002).
• Half of children may be unaware of their VI
  (Pizzarello 1998).

6
Factors associated with pediatric VI
7
Literature inadequacies

• Social epidemiological investigations
• Studies based on comprehensive eye exams
• Impairment in the context of the school district
• Direct/indirect vs. mediating determinants
• Geospatial distribution
• Relative importance of individual, school, and
  neighborhood factors

8
Primary research question

• Are there demographic, neighborhood, school, or
  school district characteristics that are most
  associated with correctable VI in school
  districts?

9
Questions and hypotheses Individual level model

• Which children have a higher prevalence of VI?
• What combination of characteristics is most
  vulnerable to VI?
• Hypothesis 1 (H1) The prevalence of VI will be
  higher in female and black, Latino, and Asian
  children compared to their counterparts.
• H2 Female black children are more likely to be
  visually impaired compared to female white
  children.

10
Questions and hypotheses Trend model

• What is the trend in VI in the past 12 years?
• If gender and racial/ethnic differences exist,
  has the gap been growing?
• H3. The rates of VI have remained similar in the
  past 12 years with slight increases.
• H4. Gender and racial/ethnic differences have
  been growing in the past 12 years. The gap in
  prevalence between white and black children is
  increasing.

11
Questions and hypotheses Neighborhood level
model

• What is the geospatial distribution of children
  with VI?
• What is the relationship of residence in
  disadvantaged areas to VI?
• H5 Children with VI are unevenly distributed
  overall in the entire area.
• H6 There are geographical clusters (hot spots)
  of visually impaired children.
• H7 There is an association between VI and
  residence in disadvantaged areas.

12
Questions and hypotheses School/district model

• Are there differences in the prevalence of VI
  based on school and school district
  characteristics?
• H8 Children in disadvantaged school districts
  will be more likely to be visually impaired
  compared to those in the opposite type of
  schools/school districts.
• H9 Children in disadvantaged schools will be
  more likely to be visually impaired compared to
  those in the opposite type of schools/school
  districts.

13
Questions and hypotheses Combined model

• When combining significant predictors (that are
  not correlated) associated with VI obtained
  through analysis based on H1-H9, what factors
  emerge as significant predictors of VI?
• How do individual, school/district, and
  neighborhood characteristics interact?
• H10 School characteristics including API and
  socioeconomic disadvantages will emerge as most
  important predictors of VI.
• H11 Students with VI who are more likely to be
  visually impaired have the following
  characteristics a.) black, in schools with poor
  academic achievement, and in poor neighborhoods,
  b.)

14
Questions and hypotheses Goodness of fit model

• Does the conceptual model specifying the relative
  relationships of factors present a good fit of
  the data?
• If not, what are other models that fit the data
  better?
• What is the strength of all predictors in
  influencing VI?
• Are they direct/indirect or mediating predictors?
• H12 The conceptual model presents a good fit
  the data. API and SES are mediating factors in
  the model and demographic factors act through the
  mediating factors but are also direct factors
  influencing impairment.

15
Methods

• Data on children examined by the UCLA Mobile Eye
  Clinic (MEC) will be analyzed
• MEC is staffed with ophthalmologists.
• Has provided exams at schools, pre-schools,
  community centers, nursing homes, and fairs since
  1989.
• Provides vouchers for free eye glasses to
  children in need.
• Provides recommendations for follow-up care.

16
Characteristics of MEC patients
17
Characteristics of school district areas where
all consenting first-graders are examined
Sources Department of Education, 2007 and
Census, 2000
18
Outcome variables

• Seven dichotomous variables
• Visual impairment visual acuity (VA) worse than
  20/40
• Blindness VA worse that 20/200
• Having myopia, hyperopia, astigmatism, refractive
  errors, and amblyopia

19
Predictors (individual and trend analysis)

• Gender (male/female)
• Race/ethnicity (white, black, Latino, Asian,
  Native American, Other)
• Year of exam
• 2000-2006 individual analysis
• 1993-2006 trend analysis

20
Predictors (neighborhood level analysis)

• Income/poverty
• Median family income
• Income in 1999 below poverty level
• Vehicles available
• Room occupancy
• Immigration status
• Being foreign born
• Language ability
• Language spoken and ability to speak English
• Living situation and parental employment

Not an exhaustive list
21
Predictors (school/district level analysis)

• Racial/ethnic distribution
• Academic success
• Academic Performance Index (API)
• Income/poverty/education
• Socioeconomic disadvantages
• Participating in free lunch program
• Average parental education level
• Language ability

Not an exhaustive list
22
Statistical analysis

• Individual, school, neighborhood and combined
• models
• Chi-square (categorical variables) and
  independent samples t (continuous variables)
  tests
• Correlations among predictors
• Logistic regression with listwise model selection
  and interactions
• Trend analysis
• Linear fit with a function of time overall and by
  race/ethnicity and gender

23
Statistical Analysis

• Goodness of fit
• Structural Equation Models (SEM)

24
SEM Example
Northouse et al 2000
25
Geospatial analysis

• Neighborhood model
• Descriptive analysis and tests of clustering

Example
Jacquez and Greiling 2003
26
Limitations

• Cross-sectional investigation
• Limited ability to generalize to all school
  districts
• Lack of direct information regarding child
  characteristics
• One tenth of children do not provide their
  race/ethnicity
• Refusal rate 5
• Eye exam Lighting not consistent when assessing
  visual acuity

27
Strengths

• Diverse population and locales in the area
• Inclusion of all consenting first-graders
• Formal assessment of vision using comprehensive
  eye exams
• Large sample size overall and by race/ethnicity
• Information about child characteristics at the
  Census tract level
• Differences among school districts and schools

28
Implications

• Social epidemiology is
• the basic science of prevention,
• serves for the development of policies that
  improve health.
• Recommend the level of resources society devotes
  to pediatric vision care
• Learning to Read Programs
• School Readiness Programs
• No Child Left Behind Act
• Vision Care for Kids Act (H.R. 507).

29
Thank you! Email gergana_at_ucla.edu
Acknowledgements E. Richard Brown Anne L.
Coleman Donald Morisky Deborah Glik Leo
Estrada Fei Yu Faye Oelrich