Get documents about "Effects of Racial and Ethnic Aggregation on Birth Outcome
Document Sample
Effects of Racial and Ethnic Aggregation
on Birth Outcome Measures for the
American Indian Population of California
(2000-2002)
David Dodds, PhD, MPH
Epidemiology and Evaluation Section
Maternal, Child and Adolescent Health Branch
California Department of Health Services
December, 2006
ddodds@dhs.ca.gov
Background: American Indian
Population and California
• California has the largest American Indian population among states:
– 333,000 (race alone, Census 2000)
– 628,000 (race alone or in combination, Census 2000)1
• California is home to:
– 12 percent of the United States population
– 13.5 percent of the American Indian/Alaska Native population2
• Los Angeles County has the largest American Indian population
among United States counties.3
• California is home to 106 of 561 federally recognized tribes.4
• Most American Indians in California declare ancestry deriving from
non-California tribes.5
*Sources: (1) Census 2000 Summary File 1, Table QT-P6; (2) State of California, Department of Finance, Census 2000:
Summary File 1 California Profile, Sacramento, California, 2002.; (3) Census 2000 PHC-T-14,Table 11; (4) Federal Register,
November 25, 2005, Vol. 70, No. 226: 71194-71198 ; (5) Census 2000 PHC-T-18, Table 19.
Background: Births and Disparities
• California ranks third in number of American Indian births
among states:
– Rank 1: Oklahoma (≈ 5,000 per year)
– Rank 2: New Mexico (≈ 3,500 per year)
– Rank 3: California (≈ 3,000 per year)1
• Disparities persist for American Indians in California and
the United States as measured by birth outcomes:
– Infant mortality
– Low birth weight
– Pre-term delivery
Source: (1) Martin, JA, et al. (2003). Births: Final Data for 2002. National Vital Statistics Reports, Vol. 52, No. 10.
Hyattsville, Maryland: National Center for Health Statistics (Table 11).
Infant Mortality Rate, California and United States,
2000-2002 Linked Files, by Race*
16.0 California
13.5 U.S.
14.0
HP 2010 (4.5)
Infant deaths per 1,000 live births
12.0 11.4
10.0 8.9
7.6
8.0 6.9
5.7 5.4
6.0 5.0
4.5 4.8
4.0
2.0
0.0
White Black American Asian or Total
Indian Pacific
Islander
*Race regardless of Hispanic ethnicity. Sources, California and United States: Matthews, TJ, et al. (2004).
Infant mortality statistics from the 2002 period linked birth/infant death data set. National Vital Statistics
Reports, Vol. 53, No. 10. Hyattsville, Maryland: National Center for Health Statistics (Table 3). Healthy People
2010, wonder.cdc.gov/data2010.
Study Problem
• Race and ethnicity are conventional categories for
assessing health disparities in public health.
• Current federal guidelines (OMB 1997) require separate
categories of race and ethnicity for data collection.1
• The federal race category “American Indian or Alaska
Native” overlaps significantly with:
– Other races (alone or in combination)
– Hispanic ethnicity
Source: (1) Office of Management and Budget, 1997, Revisions to the Standards for the Classification of Federal Data on
Race and Ethnicity, Federal Register Notice, October 30, 1997, www.whitehouse.gov/omb/fedreg/1997standards.html.
Study Questions
• Given overlap between races and ethnicity for
American Indians, do different racial/ethnic
aggregation techniques alter birth outcome
measures?
• Are some racial/ethnic aggregation techniques
more useful than others?
California: Race and Ethnicity
• California has a large Hispanic population.
• Race/ethnicity data are often aggregated before
analysis: Hispanics first into one group with remainder
groups as non-Hispanic race groups.
• This technique can diminish statistical size of the
American Indian population in states like California.
Example of “Hispanic First” Aggregation:
California Resident Live Births
by Maternal Race and Ethnicity, 2005
(N=548,700)
Hispanic: 51.5%
White: 29.8%
Asian: 11.3%
African American: 5.2%
Multiple Race: 1.3%
Pacific Islander: 0.5%
American Indian: 0.4%
All race groups are Non-Hispanic for White, Asian, African American, Multiple Race, Pacific Islander, American Indian. Data source: Birth
Statistical Master File, 2005, analysis by author.
Example Data:
California Resident Births 2000-2002 Pooled,
by Race and Ethnicity (Number)
NOT
RACE HISPANIC HISPANIC TOTAL
American Indian/Alaska Native 5,927 1,541 7,468
Asian 175,692 1,402 177,094
Black 93,070 2,018 95,088
Two or more 19,454 10,133 29,587
Native Hawaiian or Pacific Islander 7,143 234 7,377
Unknown 10,696 258 10,954
White/Other 494,077 766,457 1,260,534
TOTAL 806,059 782,043 1,588,102
Source: Linked birth and death cohort files, California, 2000-2002, analysis by author.
Example Data:
California Resident Births 2000-2002 Pooled,
by Race and Ethnicity (Percent)
RACE NOT HISPANIC HISPANIC TOTAL
American Indian/Alaska Native 0.37 0.10 0.47
Asian 11.06 0.09 11.15
Black 5.86 0.13 5.99
Two or more 1.22 0.64 1.86
Native Hawaiian or Pacific Islander 0.45 0.01 0.46
Unknown 0.67 0.02 0.69
White/Other 31.11 48.26 79.37
TOTAL 50.76 49.24 100.00
Source: Linked birth and death cohort files, California, 2000-2002, analysis by author.
Study Problem (continued)
• Among federally defined race groups in
California, overlap with Hispanic ethnicity was
highest for American Indians in Census 2000:
– 46% of Californians who chose American Indian race
alone indicated Hispanic ethnicity.1
– 39% of Californians who chose American Indian as
race alone or in combination with another race,
indicated Hispanic ethnicity.2
Source: American Fact Finder, factfinder.census.gov: (1) Census 2000 Summary File 1 (SF 1), P8. Hispanic or Latino
by Race [17] - Universe: Total population; (2) Census 2000 Summary File 1 (SF 1), P10. Hispanic or Latino by Race
(Total Races Tallied) [15] - Universe: Total races tallied.
Study Problem (continued)
• For California births, there are somewhat similar
overlaps in race and ethnicity.
• California linked files for births and deaths
during 2000-2002 show that:
– 21% of births by American Indian race alone overlap
with Hispanic ethnicity.1
– 34% of births by American Indian race alone or
combination overlap with Hispanic ethnicity.1
Source: Linked birth and death cohort files, California, 2000-2002, analysis by author.
Methods
• Test three aggregation techniques for American Indians:
– Technique 1: American Indian race alone, Not Hispanic
– Technique 2: American Indian race alone
– Technique 3: American Indian race alone or in combination
• Assess effects of aggregation technique on three birth outcome
measures:
– Infant mortality
– Percent of births with low birth weight
– Percent of pre-term births
• Data: Cohort files of linked California birth and death certificates for
pooled years 2000-2002. Three years of data:
– Improves rate stability assessed by relative standard error (RSE)
– Allows for comparison of state and national data as published by the
National Center for Health Statistics
Birth Outcome Measures with 95% Confidence Intervals:
Three Aggregation Techniques for American
Indians/Alaska Natives Resident to California, 2000-2002
Aggregation Infant Mortality Low Birth Weight Pre-Term
Technique (infant deaths (percent < 2500 Delivery
per 1,000 live grams) (percent < 37
births) wks gestation)
1) American
Indian race alone, 8.3 6.1 12.4
Not Hispanic1 (5.9, 10.6) (5.5, 6.8) (11.4, 13.4)
2) American
Indian race 7.5 6.0 12.4
alone2 (5.5, 9.5) (5.5, 6.6) (11.4, 13.2)
3) American
Indian race alone 7.5 6.9 12.3
or combination3 (6.2, 8.8) (6.5, 7.3) ( 11.7,12.8)
Birth cohort denominators, 2000-2002 pooled: 1) Infant Mortality N=5,927; LBW N=5,926, PTD N=5,365. 2) Infant Mortality N=7,468; LBW
N=7,467; PTD N=6,769. 3) Infant Mortality N=17,172; LBW N=17,172; PTD N=15,825. Data filters: LBW, birthweight 227 to 8650 grams; PTD,
gestational age 17 to 47 weeks.
Birth Outcome Measures by
Aggregation Technique for American Indians/Alaska
Natives Resident to California, 2000-2002
16.0
Race Alone, Not Hispanic
14.0 Race Alone
12.4 12.3 12.3
12.0 Race Alone or in Combination
Rate or Percent
10.0
8.3
8.0 7.5 7.5
6.9
6.1 6.0
6.0
4.0
2.0
0.0
Infant Mortality (Infant Low Birth Weight (% ) Pre-Term Delivery (% )
deaths per 1000 live births)
Source: Linked birth and death cohort files, California, 2000-2002, analysis by author.
Results
• Alternative aggregation techniques can change
descriptive birth outcome measures.
• As aggregations use less restrictive criteria:
– denominators increase in size
– confidence intervals narrow in width
– outcome measures (rates, percents) vary
• However, the three techniques tested do not produce
measures significantly different from each other.
Confidence intervals (95%) overlap across techniques
for each birth outcome measure.
Conclusions
• Technique 1: “American Indian Race Alone, Not Hispanic”
– Advantages: Captures important demographic trends in California
related to growth of the Hispanic population by immigration and high
fertility. Exclusive categories are easy to understand: all numbers sum
to 100%.
– Disadvantage: Possibly too restrictive for assessing health needs of the
California American Indian population since “race alone” population may
be moved to Hispanic
• Technique 2: “American Indian Race Alone”
– Advantages: Captures the racial heterogeneity of people with American
Indian ancestry. Avoids “loss to Hispanics.” Exclusive categories are
easy to understand: all numbers sum to 100%.
– Disadvantage: Loss of Hispanic category misses large scale
demographic trend.
• Technique 3: “American Indian Race Alone or in Combination”
– Advantage: Captures the most racial heterogeneity (i.e., racial overlap)
of people with American Indian ancestry.
– Disadvantage: Inclusive categories may be confusing because they
sum to more than 100% of the population.
Public Health Implications
• Different racial/ethnic aggregation techniques can
provide alternative perspectives for assessing health
disparities for heterogeneous populations like American
Indians of California.
• However, in the case of birth outcome measures for
California the techniques tested did not provide
significantly different measures. Other measures and
other states may be different.
• Each technique has its own advantages and
disadvantages.
Public Health Implications
• Choice of a useful racial/ethnic aggregation technique
must address purposes of data use.
• Choice of aggregation technique involves trade-offs
among criteria such as:
– Ability to identify a population for a target purpose (e.g., health
services access or detecting a particular disparity)
– Ability to compare birth outcomes measures in published
literature
– Ease of interpretation by the public
Related docs
