October 2010 Urban Institute Report on Uninsured Kids

8/8/2019 October 2010 Urban Institute Report on Uninsured Kids

1/10


2/10

Infants, and Children (WIC) to facilitate en-rollment.

The law also gave states new options to meetcitizenship requirements for child enrollees. Inaddition, it allowed states to use federal dollarsto cover legal immigrant children who had beenin the United States fewer than five years, and itprovided states with additional federal funds to

cover more children.Outcomes By February 2010, one year after

CHIPRA became law, a number of states hadeither expanded eligibility for coverage or intro-duced improvements to their enrollment andretention processes.2 By April 2010, the federalgovernment had awarded $50 million in out-reach grants, including $40 million to organiza-tions in forty-two states and an additional$10 million for targeting Native American chil-dren. These policy changes are expected tochange the composition of the population ofchildren enrolled in public coverage and raise

participation rates among children who are al-ready eligible.

At the same time, however, the ongoing reces-sion and state budget shortfalls could reducestate-level efforts to promote greater enrollmentand retention among eligible children. Ulti-mately, policy changes to be implemented in2014 under the Patient Protection and Afford-able Care Act of 2010 will introduce majorchanges to Medicaid and CHIP coverage for chil-dren and parents.

New Data Source To assess progress coveringuninsured children under CHIPRA and, ulti-

mately, through health reform, information onnationwide participation and coverage isneeded. Data from the state and local levels,and for different subgroups of children, are alsoneeded. Research suggests that participation inMedicaid/CHIPthat is,the ratio of eligible chil-dren enrolled in the programs to that number ofchildren plusuninsuredeligiblechildrenvariesacross regions of the country.8 To date, however,it has not been possible to produce robust esti-mates foreachstate, because of limitations in theavailable data.

The situation has now changed. The American

Community Survey, which includes an annualsample of approximately 700,000 children na-tionwide, began asking about health insurancecoverage in 2008.11 The availability of health in-surance data from this survey, along with accom-panying household and income data, allowedus to develop a Medicaid/CHIP eligibility modelfor the survey that produces more precise stateand local estimates than had been possiblepreviously.

New Estimates This paper provides nationaland state-level estimates of Medicaid/CHIP par-

ticipation rates and of the number of uninsuredchildrenwho areeligiblefor the programs,basedon 2008 data from the American CommunitySurvey.We found substantial variation in partici-pation rates across states and among subgroupsof children.We also foundthat themajority of un-insured children who are eligible for Medicaid/CHIP butnot enrolledare concentratedin a small

number of the most populous states.

Study Data And MethodsData Source These estimates are derived fromthe 2008 American Community Survey, an an-nual survey fielded continuously over a twelve-month period by the U.S. Census Bureau.12 Weused an augmented version of the survey pre-pared by the University of MinnesotaPopulationCenter.13 The survey has a reported response rateof 97.9 percent (range: 91.4 percent in Washing-ton, D.C., to 99.4 percent in Wisconsin).14

The survey uses an area frame that includeshouseholds with and without telephones (land-line or mobile). A mixed-mode survey, it startswith a mail-back questionnaire (56.6 percent ofthe sample is completed by mail). Nonrespond-ers receive a follow-up phone call, and a subsam-ple of remaining nonresponders is interviewedin person.15 The estimates presented here focuson children age eighteen and younger in thecivilian noninstitutionalized population, includ-ing college students in dorms and a small num-ber of other children living in group quarters,such as residential treatment facilities.

In 2008 a question was added to the surveythat asked about coverage status in differenttypes of insurance plans for each individual inthe household at the time of the survey(Exhibit 1). We classified children as uninsuredif they did not have coverage under categories Athrough F (including those recoded from thewrite-in option, category H) and if they werenot classified as having coverage based on otherinformation collected in the survey.1618 (See the

Appendix for a description of further adjust-ments made to the coverage data.)19 Becausethe data are collected continuously over a

twelve-month period, the coverage estimates re-present an average day in the calendar year.

Research suggests that with the exception ofdirect-purchase coverage, the survey coverageestimates are generally valid, with estimatesabout the same for other coverage categoriesas thosefrom two other federallyfunded surveys,the Current Population Survey and the NationalHealth Interview Survey.16 However, there is con-cern that the survey may understate Medicaidand CHIP coverage because, in addition to theknown underreporting of public coverage on

Web Fir st

2 H e a lt h A f f a i r s O c to b er 2 0 1 0 2 9 : 10


3/10

household surveys, the American Community

Surveyunlike the other surveys cited abovedoes not specifically mention CHIP or give re-spondents names for their states particularMedicaid and CHIP programs.

Adjustments To address the underreportingof Medicaid and CHIP, we made adjustmentsbased on approaches that have been applied toother surveys.18,20We applied a set of logical editrules that the Census Bureau has developed forthis survey, as well as additional edits that takeadvantage of other information collected in thesurvey (see the Appendix).19 The edits reducedthe estimated number of uninsured children in

the survey from 8.2 million to 7.3 million, whichis slightly lower than the National Health Inter-

view Survey estimate of 7.4 million uninsuredchildren for the same period.21 Our edits in-creased the estimated number of children withMedicaid/CHIP as their primary coverage byroughly 4.4 million to a level that is just about7 percent lower thanthe comparable administra-tive count for June 2008.22

Analytic Model This analysis relied on theUrban Institute Health Policy Centers ACSMedicaid/CHIP Eligibility Simulation Model,which builds on the model developed for the

Current Population Survey by Lisa Dubay and Allison Cook.23,24 Despite the differences be-tween surveys, the American Community Surveyeligibility simulation for 2008 appears robustcompared to that developed for the Current Pop-ulation Survey. For example, the numbers andcharacteristics of children according to their eli-gibility for Medicaid/CHIP and their pathway toeligibilitythat is, Medicaid versus CHIParesimilar for both surveys.25

We defined participation rates as the ratio ofeligible children enrolled in Medicaid/CHIP to

those children plus uninsured children who are

eligible for Medicaid/CHIP. In our core esti-mates, we excluded from these counts childrenenrolled in one of the programs who also haveother coverage (such as private or military cover-age) and those with Medicaid/CHIP coveragewho do not have a known eligibility pathway.26

We do not currently have a method for includingin the denominator of the participation ratethose uninsured children who may similarly beeligible for Medicaid/CHIP coverage but dontappear to be, based on the information that isavailable on the survey.

Variation in participation within states can be

assessed using Public Use Microdata Areas.These areas, numbering more than 2,000, areusually defined in terms of counties: A microdataarea can cover a single county, a combination ofwhole counties, or a part of a large county. Esti-mates are presented by census region and divi-sion, age, race or ethnicity, citizenship status,presence of English-speaking parents, family in-come (defined in terms of the childs health in-surance unit), participation in the SupplementalNutrition Assistance Program (formerly knownas food stamps) at some point in the prior twelvemonths, and the presence of a telephone (either

landline or mobile) in the household. All estimates used weights provided by the

Census Bureau. Standard errors were calculatedusing replicate weights that took into accountthe complex nature of the sample design.

Study FindingsCharacteristics Of Uninsured Children Ac-cording to our revised coverage estimates, some7.3 million children were uninsured on an aver-age day in 2008, of whom 4.7 million (65 per-

EXHIBIT 1

American Community Survey Question On Health Insurance

Is this person CURRENTLY covered by any of the following types of health insurance or health coverage plans? MarkYes or No for EACH type of coverage in items AH.

A. Insurance through a current or former employer or union (of this person or another family member)

B. Insurance purchased directly from an insurance company (of this person or another family member)

C. Medicare, for people age 65 and over, or people with certain disabilities

D. Medicaid, Medical Assistance, or any kind of government-assistance plan for those with low incomes or a disabilityE. TRICARE or other military health care

F. VA (including those who have ever enrolled in or used VA health care)

G. Indian Health Service

H. Any other type of health insurance or health coverage planspecify

Source U.S. Census Bureau, American Community Survey, 2008. Notes TRICARE is the program formerly known as CHAMPUS; itcovers uniformed service members, retirees, and their families worldwide. VA is Department of Veterans Affairs.

O c to b er 2 0 1 0 2 9 : 10 H e a lt h A f f a i r s 3


4/10

cent) were eligible for Medicaid or CHIP but notenrolled (Exhibit 2).26 Of these 4.7 million chil-dren, 3 million had family incomes below133 percent of the federal poverty level,27 1.2 mil-lion had family incomes of 133200 percent ofpoverty, and 500,000 had incomes above200 percent of poverty.

Although 64 percent of uninsured children

eligible for Medicaid/CHIP had family incomesbelow 133 percent of poverty, children in thisincome group participated in Medicaid/CHIPat higher rates relative to higher-income chil-dren (see the discussion below). Most of theremaining 2.5 million uninsured children didnot qualify for Medicaid/CHIP because theirfamily incomes exceeded income eligibilitythresholds in 2008.

Patterns Among States The number of un-insured children who wereeligible for Medicaid/CHIP but not enrolled was heavily concentratedin a relatively small number of populous states

(Exhibit 3). Just three states combinedCalifor-nia, Texas, and Floridacontained 38.6 percentof all eligible uninsured children in the country.Moreover, an estimated 61 percent (about2.9 million) of all eligible uninsured childrenlived in the ten states shown in the exhibit.

The main reason these states accounted forsuch a large share of the eligible uninsured chil-dren is that they also contained a disproportion-

ate share of children: 52 percent of all childrenand 56 percent of eligible children in the nation(data not shown). However, Florida, Texas, and

Arizona also had participation rates that werewell below the national average (Exhibit 4)69.8 percent, 74.7 percent, and 76.6 percent,respectively.

Participation Rates Overall, we estimated

that the national rate of Medicaid/CHIP partici-pation for children was 81.8 percent in 2008(Exhibit 4).28 The median rate across stateswas even higher, 83.3 percent (AppendixExhibit A.1).19Although not exactly comparable,the Medicaid/CHIP participation rate we esti-mated for children was much higher than theparticipation rates typically found in othergovernment programs. This is probably due toconcerted efforts to improve Medicaid/CHIP eli-gibility,enrollment and retention processes,andoutreach.3,29We found higher participation ratesfor states in the Northeast (87.7 percent) and

Midwest (85.3 percent) census regions andlower rates for states in the West (78.8 percent)and South (79.8 percent) regions (Exhibit 5).STATE VARIATIONS: Participation rates var-

ied greatly across states, from lows of 55.4 per-centin Nevadaand 66.2 percent in Utahto highsof 95.4 percent and 95.2 percent inthe District ofColumbia and Massachusetts, respectively.Hawaii, Maine, Massachusetts, Vermont, andthe District of Columbia had participation ratesof 91 percent or higher, andArkansas, Kentucky,Louisiana, Michigan, New York, and West Vir-ginia had rates of 8890 percent. A total of thir-

teen states had participation rates under80 percent (Alaska, Arizona, Colorado, Idaho,Florida, Montana, Nevada, North Dakota,Oregon, South Carolina, Texas, Utah, and

Wyoming).Participation rates for the states in the top

quintile, or 20 percent of states, ranged from88.4 percent to 95.4 percent, while participationrates for the states in the bottom quintile rangedfrom 55.4 percent to 75.2 percent (AppendixExhibit A.1).19 The top quintile included statesof different sizes, racial and ethnic composition,and income levels (data not shown), and at least

one state from each census region. However,neither the Mountainnor the WestNorth Centralcensus division had a state in the top quintile.Five of the six states in the New England censusdivision were in the top two quintiles, and threewere in the top quintile.

Seven of the ten states with participation ratesin the lowest quintile were in the West censusregion, andfive of these stateswere in theMoun-tain division. All eight states in that division hadparticipation rates in the bottom two quintiles.No state in the lowest quintile was in the Middle

EXHIBIT 2

Eligibility Of Uninsured Children For Coverage Through Medicaid Or The Childrens HealthInsurance Program (CHIP), 2008

Source Authors analysis using the Urban Institute Health Policy Center s ACS Medicaid/CHIP Elig-ibility Simulation Model, based on data from the 2008 American Community Survey. Notes Esti-mates reflect an adjustment for the underreporting of Medicaid/CHIP coverage on the survey. Ofthe 7.3 million uninsured U.S. children, 4.7 million are eligible for Medicaid or CHIP. Percentagesshown here are percentages of the 7.3 million children. Cumulative percentages might not total100 because of rounding.

Web Fir st



5/10

Atlantic, New England, East North Central, orEast South Central division.I N DI A N H EA LT H S ER VI C E: As indicated

above, state-level participation rates treated asuninsured all children reporting Indian HealthService but no other sourceof coverage.Whenwerecalculated the participation rates by not count-ing these children as uninsured, the national

participation rate remained the same. However,the estimated rates rose by three percentagepoints or more in six states (Alaska, Montana,New Mexico, North Dakota, Oklahoma, andSouth Dakota), with the largest increase(14.5 percentage points) occurring in Alaska(data not shown).

Thus, the lower Medicaid/CHIP participationrates in some western states may be linked to therelatively larger shares of Native American chil-dren in those states. New higher federal match-ingratesarenowavailabletostatestocoverthesechildren, whichmay increasetheir enrollment in

Medicaid and CHIP. V ARI ATI O N W I T H I N S T AT E S: Participation

in the programs also appeared to vary consider-ably across areas within a given state (data notshown). For example, rates in Florida, a state inthe bottom quintile of participation, variedacross areas in the state from a low of 45.8 per-cent to a high of 92.3 percent. V ARI ATI O N AC RO S S S UB G RO UP S: We also

foundsubstantial variation in participation ratesacross different subgroups of children who wereeligible for coverage (Exhibit 6). Overall, adoles-cents were aboutsevento nine percentage points

less likely thanyounger children to participate inMedicaid/CHIP. African American children andthose in the other/multiple race category hadthe highest participation rates among eligiblechildren, at around 87 percent; Hispanic and

Asian/Pacific Islander children had participa-tion rates of 78.8 percent and 79.4 percent, re-spectively.

Although American Indian/Alaska Native chil-dren appeared to have the lowest participationrates (68.0percent), their participation rateroseto 91.6 percent if we did not treat as uninsuredthe children reported as covered only through

the Indian Health Service.We found higher par-ticipation rates among citizen children with citi-zen parents and among those who had at leastone English-speaking parent thanamong citizenchildren with no parents who were citizens orwith no English-speaking parent.

Participation rates were higher among chil-dren with family incomes below 133 percent ofpoverty (84.2 percent) and among those whowere in households that received help throughthe Supplemental Nutrition Assistance Program(formerly known as food stamps; 93.5 percent),

compared to higher-income children and chil-dren not living in households receiving foodstamps.

Despite theirrelatively highparticipation rate,the poorest children made up a sizable majority(63.4 percent) of all children who were eligiblebut uninsured. Children living in homes withoutphones had participation rates that were almost

ten percentage points lower than those of chil-dren who had phones in their homes.

As seen in Exhibit 6, the remaining eligibleuninsured children were heterogeneous alonga number of different dimensions. For example,39.1 percent were Hispanic; 36.9 percent werewhite; 15.8 percent were black; and the remain-ing 8.2 percent included Asian/Pacific Islanders,

American Indian/Alaska Natives, and childrenin the other/multiple race category.

In addition, although the majority did not livein households that received food stamps, Ex-press Lane strategies that connect families

who receive food stamps to Medicaid and CHIPcoverage could help reduce uninsurance amongthe 15.4 percent of uninsured children who wereeligible for Medicaid/CHIP and whose familiesdid receive food stamps.

The participation patterns we found at the na-tional level generally also appeared in the tenstates with the largest number of eligible unin-sured children. In other words, we generallyfound lower participation rates for adolescentsand higher-income children; for children whohad no parents who were citizens or spoke En-

EXHIBIT 3

Number Of Eligible Uninsured Children In Selected States, 2008

State

Distribution of eligible uninsuredchildren

Cumulative share of totalU.S. uninsured (%)Number Percent 90% CI

United States 4,708,000

California 695,000 14.8 14.215.3 14.8

Texas 693,000 14.7 14.315.1 29.5

Florida 429,000 9.1 8.89.4 38.6

Georgia 193,000 4.1 3.84.4 42.7

New York 164,000 3.5 3.33.7 46.2

Arizona 155,000 3.3 3.13.5 49.5

Illinois 146,000 3.1 2.83.4 52.6

Ohio 144,000 3.1 2.83.3 55.6

Pennsylvania 129,000 2.7 2.53.0 58.4

North Carolina 125,000 2.7 2.42.9 61.0

Source Authors analysis using the Urban Institute Health Policy Centers ACS Medicaid/CHIPEligibility Simulation Model, based on data from the 2008 American Community Survey. Notes

Estimates of uninsured children reflect an adjustment for the underreporting of Medicaid andChildrens Health Insurance Program (CHIP) coverage on the survey. Numbers of uninsuredchildren are rounded to the nearest thousand. Cumulative percentages might not total 100because of rounding. CI is confidence interval.

O c to b er 2 0 1 0 2 9 : 10 H e a lt h A f f a i r s 5


6/10

glish; and for those without phone access (seeAppendix Exhibits A.3A.12).19 In contrast, thecomposition of eligible uninsured children ap-peared to vary across states, particularly withrespect to race and ethnicity, citizenship status,and language.

DiscussionThese new estimates suggest that as of 2008,

nearly five million uninsured children were eli-gible for but not enrolled in Medicaid/CHIP. Toachieve the goal of reaching and enrolling all ofthese children, as set forth by the HHS secretary,progressis needed in allstates. However, it is notclear how much higher participation can be inthe states that already have rates greater than90 percent, given the dynamic nature of familycircumstances andeligibility for public coverage.

Absent increases in Medicaid/CHIP participa-tion in the ten states that account for 61 percentof all eligible uninsured children, there would

still be close to three million uninsured childrennationally who are eligible for Medicaid/CHIPeven if the remaining forty states were able toachieve participation rates close to 100 percent.Moreover, because California, Florida, andTexastogether account for 1.8 million of the total eli-gible uninsured children, increasing participa-tion in those three states will be critical toreaching the national goal.

Policy Implications Theseestimates indicate

that outreachefforts andpolicy reforms aimed atimproving eligibility, enrollment, and retentionprocesses will need to reach children of differentages,incomes, races, ethnic groups, and primarylanguage, given the diversity of the remainingeligible uninsured population. At the same time,however, targeted enrollment, retention, andoutreach efforts may be needed for children inparticular subgroups who constitute a dispro-portionate share of the eligible uninsured chil-dren nationally and in individual states.

Research suggests that some policies are par-

EXHIBIT 4

Childrens Participation In Medicaid And The Childrens Health Insurance Program (CHIP), By State, 2008

Statisticallyhigher thannational rate

Statisticallylower thannational rate

Source Authors analysis using the Urban Institute Health Policy Centers ACS Medicaid/CHIP Eligibility Simulation Model, based on data from the 2008 AmericanCommunity Survey. Notes Estimates reflect an adjustment for the underreporting of Medicaid/CHIP coverage on the survey. The national participation rate in 2008

was 81.8 percent. Statistical significance denotes difference from the national percentage at p < 0:10 level.

Web Fir st



7/10

ticularly promising, such as using income taxdata or data from other means-tested programsto automatically qualify children for Medicaidand CHIP, which could reach the vast majorityof eligible uninsured children.30,31

As indicated above, the estimated nationalMedicaid/CHIP participationrate is 81.8 percent

for children. Our data indicate that ten stateshave participation rates very close to or above90 percent. Because these states constitute a di-

verse group in terms of their size, income dis-tribution, racial and ethnic composition, andregion, it seems clear that high participationrates can be achieved across a range of differentcircumstances.

The data raise questions about the underlyingreasons for the observed state-level variation inparticipationrates. For example, it is not obviouswhyso many western states, particularlythoseinthe Census Bureaus Mountain division, have

participation rates in the lowest quintile, orwhy such a large share of New England stateshave rates in the top quintile.

Questions For Further Research Two re-lated questions need more research. First, weneed to understand how much of the variationin participation across statescan be explainedbydifferences in the characteristics of states relatedto such factors as population density, per capitaincome, political culture, racial or ethnic com-position, and access to employer coverage. Sec-ond, we need to know how much is related to

specific state policies regarding Medicaid/CHIPincome eligibility thresholds for both childrenand their parents; Medicaid and CHIP outreachefforts; and eligibility determination, enroll-ment, and retention. Analysis that is in processcould shed light on the reasons why participa-tion rates vary so much both within and across

states.Study Limitations Our analysishas a number

of limitations. First, despite our attempts to pro-duce reliable coverage estimates with the Ameri-can Community Survey, there still may bemeasurement errors, which could introduce biasinto our estimates of eligible uninsured childrenand the participation rates at the national, state,within-state, and subgroup levels.

Although our national estimatesmatch closelythose derived from other surveys, targeted meth-odological research on theAmerican CommunitySurvey coverage estimates would provide much

more certainty about our adjustments and a bet-ter understanding of the response patterns, par-ticularly with respect to the underreporting ofMedicaid/CHIP coverage. Such research couldinvolve a number of steps, including matchesof the survey to Medicaid and CHIP enrollmentrecords, more cognitive testing to understandhow low-income families and respondents withlow educational attainment or whose primarylanguage is not English interpret the coveragecategories included on the survey.

Also needed are thorough follow-up surveys

EXHIBIT 5

Rates Of Childrens Participation In Medicaid And The Childrens Health Insurance Program (CHIP), By Census Region andDivision, 2008

Region or division

Medicaid/CHIP participation rate

Percent 90% CI

United States 81.8 81.682.1

Northeast region 87.7* 87.288.1New England divisiona 90.6* 89.791.5Middle Atlantic divisionb 86.8* 86.287.4

Midwest region 85.3* 84.885.8East North Central divisionc 86.1* 85.586.7West North Central divisiond 83.2* 82.384.2

South region 79.8* 79.480.2South Atlantic divisione 78.9* 78.379.6East South Central divisionf 85.7* 85.086.4West South Central divisiong 78.3a 77.778.9

West region 78.8* 78.379.3Mountain divisionh 72.4* 71.373.4Pacific divisionj 81.4 80.882.0

Source Authors analysis using the Urban Institute Health Policy Center s ACS Medicaid/CHIP Eligibility Simulation Model, based ondata from the 2008 American Community Survey. Notes Estimates reflect an adjustment for the underreporting of Medicaid/CHIPcoverage on the survey. Statistical significance denotes difference from the national percentage. CI is confidence interval.

a CT, ME, MA, NH, RI, VT. b NJ, NY, PA. c IN, IL, MI, OH, WI. d IA, KS, MN, MO, NE, ND, SD. e DE, DC, FL, GA, MD, NC, SC, VA, WV.f AL, KY, MS, TN. g AR, LA, OK, TX. h AZ, CO, ID, MT, NM, NV, UT, WY. j AK, CA, HI, OR, WA. *p < 0:10

O c to b er 2 0 1 0 2 9 : 1 0 H e a lt h A f f a i r s 7


8/10

with a sample of families who reported on thesurvey thattheir childrenhad either no coverage,

employer-sponsored insurance, or nongroupcoveragebut whoappear to be enrolledin Medic-aid or CHIPbased on other information providedon thesurvey. It maybe particularlyimportant toexamine these factors in states where there is alarger apparent gap between the surveys totalsof children in Medicaid/CHIP and the enroll-ment totals from administrative data.

Second, our eligibility simulation also hasmeasurement error, particularly with respectto noncitizen children, which could introducebias as well. Third, estimates for smaller states

(such as North Dakota, Vermont, and Wyoming)and the District of Columbia are less precise be-

cause of the relatively smaller samples availablefor them in the public use files. It will be impor-tant to assess how robust these estimates are,using the full sample, additional analyses, andanother year of data.

Recent Legislative Changes Our analysispertains to 2008, before CHIPRA was enacted.Since then, a number of important changes haveoccurred that could affect the estimates. In par-ticular, a number of states have expanded eli-gibility for public coverage or introducedpolicies aimed at increasing enrollment and re-

EXHIBIT 6

Rates Of Childrens Participation In Medicaid And The Childrens Health Insurance Program (CHIP) And The Distribution OfEligible Uninsured Children, By Child And Household Characteristics, 2008

CharacteristicMedicaid/CHIP participationrate (%)

Distribution of eligible uninsuredchildren

Percent 90% CITotal 81.8 100.0

Age (years)05 85.5* 30.1 29.630.6612 82.6* 33.5 33.033.91318a 75.9 36.4 35.837.0

Ethnicity or race

Hispanic 78.8* 39.1 38.439.8Whitea 81.8 36.9 36.237.7Black/African American 86.8* 15.8 15.216.3Asian/Pacific Islander 79.4* 3.2 3.03.3American Indian/Alaskan Native 68.0* 2.2 2.12.4Other/multiple races 86.7* 2.8 2.63.0

Citizenship status

Citizen child with no citizen parents 78.5* 17.3 16.717.8Citizen child with citizen parentsa 83.6 65.9 65.266.5

Noncitizen child 69.1* 4.6 4.34.8Child not living with parents 76.6* 12.3 11.912.7

English-speaking parent in home

At least onea 83.1 73.5 72.974.0None 77.2* 14.2 13.814.7Child not living with parents 76.6* 12.3 11.912.7

Family Income (as percent of poverty)

0132%a 84.2 63.4 62.764.2133%199% 76.0* 25.8 25.126.6200% or more 73.5* 10.7 10.311.2

Household receives food stamps

No 72.9* 84.6 84.085.1Yesa 93.5 15.4 14.916.0

Access to phone in home

Noa 72.9 5.3 4.95.7Yes 82.2* 94.7 94.395.1

Source Authors analysis using the Urban Institute Health Policy Center s ACS Medicaid/CHIP Eligibility Simulation Model, based ondata from the 2008 American Community Survey. Notes Estimates reflect an adjustment for the underreporting of Medicaid/CHIPcoverage on the survey. Statistical significance denotes difference from the reference group in each category. Cumulative percentagesmight not total 100 because of rounding. See Appendix Exhibit A.2 for the confidence intervals of the estimates provided (theAppendix can be accessed by clicking the Appendix link in the box to the right of the article online). Family income is defined interms of the childs health insurance unit. aReference group. *p < 0:10

Web Fir st



9/10


10/10

Suitland (MD): Census Bureau;2009 [cited 2010 Aug 16]. Availablefrom: http://www.census.gov/acs/www/UseData/sse/

15 Griffin D, Hughes T. Mixed modedata collection in the AmericanCommunity Survey. Paper presentedat: American Association of PublicOpinion Research Conference. 2010May 1316; Chicago, IL.

16 Turner J, Boudreaux M, Lynch V. Apreliminary evaluation of health in-surance coverage in the 2008

American Community Survey [In-ternet]. Suitland (MD): U.S. CensusBureau; 2009 Sep 22 [cited 2010

Aug 16]. Available from: http://www.census.gov/hhes/www/hlthins/data/acs/2008/2008

ACS_healthins.pdf17 The Indian Health Service (IHS) is

not typically counted as health in-surance coverage because of limita-tions in the scope of availableservices and geographic reach of IHSfacilities. In 2008, approximately140,000 children were estimated to

have only IHS coverage.18 Lynch V, Boudreaux M, Davern M.

Applying and evaluating logicalcoverage edits to health insurancecoverage in the American Commu-nity Survey [Internet]. Suitland(MD): U.S. Census Bureau; 2010 Jul[cited 2010 Aug 16]. Available from:http://www.census.gov/hhes/www/hlthins/publications/coverage_edits_final.pdf

19 The Appendix can be accessed byclicking the Appendix link in the boxto the right of the article online.

20 DavernM, Klerman JA,ZiegenfussJ,Lynch V, Baugh D, Greenberg G. Apartially corrected estimate ofMedicaid enrollment and uninsur-ance: results from an imputationalmodel developed off linked surveyand administrative data. J Econ SocMeas. 2009;34(4):21940.

21 According to Urban Institute tabu-lations of data from the 2008 Na-tional Health Interview Survey.

22 The American Community Surveyincludes 1.4 million children withemployer-sponsored or other cover-age who were also reported to haveMedicaid/CHIP at the time of thesurvey. See the Appendix (as inNote 19) for information on how theMedicaid and CHIP administrative

totals were derived for June 2008.The difference between our adjustedstate Medicaid/CHIP total and therespective administrative total variesacross states; for more than half of

the states, the difference is less than10 percent.

23 Dubay L, Cook A. How will the un-insured be affected by health re-form? Washington (DC): KaiserCommission on Medicaid and theUninsured; 2009 Aug.

24 See the Appendix, as in Note 19, formore information on the ACS eli-gibility simulation.

25 Estimates of eligible unenrolledchildren were similar for the CurrentPopulation Survey, which found thatan estimated 4.9 million (68 per-cent) uninsured children were eli-gible for Medicaid or CHIP in 2009.Urban Institute tabulations usingthe Health Policy Centers CPSMedicaid/CHIP Eligibility Model,based on data from the 2009 CPS.

26 We also produced additional sets ofparticipation estimates to show theimpact of excluding children whoappear to have Medicaid as well asother coverage (so-called wrap-around coverage) and of excludingchildren notfound to be eligible. The

pattern of findings reported here isconsistent with these additional es-timates in terms of variation acrossstates and subgroups. For example,participation rates for eligible chil-dren that included children withother coverage were higher thanparticipation rates that excludedthese children (82.6 percent com-pared to 81.8 percent). However,state-level patterns remained con-sistent: Nevada and Utah were thelowest (57.5 percent and 67.9 per-cent) and Massachusetts and theDistrict of Columbia were the high-est (95.6 percent and 95.5 percent).Participation rates that includedchildren reported to have Medicaid/CHIP but not found to be eligible inthe model were higher than thosethat excluded them, at around84.0 percent; however, again, thestate-level patterns of participationwere similar to those among eligiblechildren with no dual report ofMedicaid/CHIP and employer-spon-sored insurance; Nevada and Utahwere again the lowest (62.7 percentand 73.4 percent) and Massachusettsand the District of Columbia, thehighest (95.9 percent and95.6 percent).

27 We usedCensusBureauguidelines to

define poverty levels for descriptivestatistics.28 The national participation estimate

for the 2009 Current PopulationSurvey when including children re-

ported to have both Medicaid andother coverage was 81.7 percent,compared to 82.6 percent for the

American Community Survey. UrbanInstitute tabulations using theHealth Policy Centers CPS Medic-aid/CHIP Eligibility Model, based ondata from the 2009 CPS.

29 Making valid comparisons of par-ticipation rates between Medicaid/

CHIP and other programs is difficultbecause of differences in how thetarget population is defined acrossprograms, particularly as it relates toexcluding from the denominatorthose Medicaid/CHIP-eligible chil-dren who had private coverage. U.S.Government Accountability Office.Means tested programs: informationon program access can be an im-portant management tool [Internet].

Washington (DC): GAO; 2005 Mar[cited 2010 Aug 16]. Available from:http://www.gao.gov/new.items/d05221.pdf

30 Dorn S, Kenney GM. Automaticallyenrolling eligible children and fam-

ilies into Medicaid and SCHIP: op-portunities, obstacles, and optionsfor federal policymakers [Internet].New York (NY): CommonwealthFund; 2006 Jun [cited 2010 Aug 16].

Available from: http://mobile.commonwealthfund.org/~/media/Files/Publications/Fund%20Report/2006/Jun/Automatically%20Enrolling%20Eligible%20Children%20and%20Families%20Into%20Medicaid%20and%20SCHIP%20%20Opportunities%20%20Obsta/Dorn_auto%20enrollingchildren_931%20pdf.pdf

31 Dorn S, Garrett B, Perry C, Clemans-Cope L, Lucas A. Nine in ten: usingthe tax system to enroll eligible,uninsured children into Medicaidand SCHIP [Internet]. Washington(DC): Urban Institute; 2009 Jan[cited 2010 Aug 16]. Available from:http://www.urban.org/UploadedPDF/411844_tax_system.pdf

32 The following item includes a dis-cussion of enhancements to federalsurveys that could strengthen theirability to produce reliable data formonitoring coverage changes underboth CHIPRA and the AffordableCare Act. Kenney G, Lynch V. Mon-itoring childrens health insurance

coverage under CHIPRA usingfederal surveys. Washington (DC):National Academies Press; 2010 Jul.

Web Fir st

10 H e a lt h A f f a i r s O c to b er 2 0 1 0 2 9 : 1 0

October 2010 Urban Institute Report on Uninsured Kids

Documents

Transcript of October 2010 Urban Institute Report on Uninsured Kids