Trends in Type of Health Insurance Coverage for US Children and

Their Parents, 19982011

Jennifer E. DeVoe, MD, DPhil; Carrie J. Tillotson, MPH; Miguel Marino, PhD;
Jean OMalley, MPH; Heather Angier, MPH; Lorraine S. Wallace, PhD;
Rachel Gold, PhD, MPH
From the Department of Family Medicine (Dr DeVoe, Dr Marino, and Ms Angier), Division of
Biostatistics, Department of Public Health and Preventive Medicine (Ms Tillotson, Dr
Marino, and Ms OMalley), Oregon Health & Science University, Center for Health
Research, Kaiser Permanente Northwest (Dr Gold), Portland, Ore; and Department of
Family Medicine, The Ohio State University, Columbus, Ohio (Dr Wallace)
The authors declare that they have no conflict of interest.
Address correspondence to Heather Angier, MPH, Department of Family Medicine, Oregon
Health & Science University, 3181 Sam Jackson
Rd, Mailcode FM, Portland, OR 97239 (email: angierh@ohsu.edu). Received for
publication January 7, 2015; accepted June
15, 2015.
ABSTRACT
OBJECTIVE: To examine trends in health insurance type among
US children and their parents.
METHODS: Using the Medical Expenditure Panel Survey (19982011), we linked each child (n
120,521; weighted n z 70 million) with his or her parent or parents and assessed
patterns of full-year health insurance type, stratified by income. We examined longitudinal
insurance trends using joinpoint regression and further explored these trends with adjusted
regression models.
RESULTS: When comparing 1998 to 2011, the percentage of
low-income families with both child and parent or parents pri- vately insured decreased from 29.2%
to 19.1%, with an esti- mated decline of 0.86 (95% confidence interval, 1.10,
0.63) unadjusted percentage points per year; middle-income families experienced a drop from
74.5% to 66.3%, a yearly un- adjusted percentage point decrease of 0.73 (95% confidence
interval,
0.98,
0.48). The discordant pattern of publicly insured children with uninsured
parents increased from 10.4% to 27.2% among low-income families and from 1.4% to 6.7% among
middle-income families. Results from adjusted models were similar to joinpoint regression findings.
CONCLUSIONS: During the past decade, low- and middleincome US families experienced a decrease in the percentage of childparent pairs with private
health insurance and pairs without insurance. Concurrently, there was a rise in discordant coverage
patternsmainly publicly insured children with unin- sured parents.
KEYWORDS: access to care; family health; health insurance;
uninsured
ACADEMIC PEDIATRICS 2015;-:18
WHATS NEW
Trends in health insurance type have changed over the past decade for low- and middle-income
US families: private coverage and uninsurance have decreased, while discordance between parent
and child coverage has increased.
STABLE HEALTH
INSURANCE leads to better access to health care services and improved
health outcomes.14
Over the past decades, political and economic changes have affected access to and affordability of
coverage for families in the United Statesnotably, private health insurance costs have seen steep
increases. Though some families obtained coverage for their children through expansions in the

Childrens Health Insurance Program (CHIP), few public coverage options existed for adults (aged
1964 years) before 2014.5,6
Parental coverage status has an independent effect on childrens health insurance and access to
care, regardless of the childs coverage status.79 Previous research
utilizing a natural experiment that randomized adults to coverage found a causal link between parent
and child health insurance status.10 Thus, it is important to consider trends in childrens health
insurance coverage in conjunc- tion with trends affecting parents. Most past studies of health
insurance have focused on adults or children sepa- rately; those that considered both children and
parents did not assess type of coverage.714 To address this gap in the literature, we examined
full-year patterns of family health insurance coverage type among US children and their parents for
1998 through 2011, stratified by income.
METHODS
DATA SOURCE AND STUDY POPULATION
We analyzed data from 1998 through 2011 of the Med- ical Expenditure Panel SurveyHousehold
Component (MEPS-HC).15 MEPS-HC respondents are interviewed 5 times over a 2-year period,
with an overlapping panel design; annual public use files contain data from a single year for 2
consecutive panels. Each year of data constitutes
ACADEMIC PEDIATRICS

2
DEVOE ET AL
ACADEMIC PEDIATRICS
a nationally representative sample. Details about the
MEPS-HC are available elsewhere.15,16
The study population included children aged 0 to 17 years, with responses to at least 1 full year of
the survey (n 126,093). We linked each child with a parent or parents in the same household to
construct childparent pairs. We excluded children for whom no identifiable parent records could be
linked (n 4048), and for whom insurance infor- mation for the child or parent was not available for
the full year (n 1524). Our final sample size was 120,521 children, weighted to represent a yearly
average of approximately 70 million children in the civilian, noninstitutionalized US pop- ulation.
CONSTRUCTING HEALTH INSURANCE TYPE VARIABLES
The MEPS-HC contains variables for whether a person had health insurance for at least 1 day in
each calendar month of each year, and whether it was public or private insurance. Using these, we
constructed variables represent- ing full-year health insurance type, classified as: 1) having private
insurance if a person had insurance in 12 months of the year of which 1 or more months included
private insur- ance (those with a combination of public and private insur- ance were included in this
category); 2) having public insurance if a person was insured in 12 months of the year and had
public insurance only; and 3) being unin- sured, in which the person was reported as having no insurance in 1 or more months of the year. We included those with a combination of public and private
insurance in the private category to match MEPS-HC health insurance var- iable categorization15;
we considered those who did not have insurance in 1 or more months of a given year as uninsured because previous research has shown that preventive service rates for patients with partial
health insurance are different from those with full-year coverage17 and are similar to those with
no coverage.18,19
We then created a combined variable that paired full- year health insurance type for a child with
that of his or her parent. We grouped child and parent type of health in- surance into 9 mutually
exclusive categories (child type/ parent type): private/private; private/public, private/unin- sured;
public/private; public/public; public/uninsured; uninsured/private; uninsured/public; and
uninsured/unin- sured. In cases where a child had 2 parents linked, parent insurance was considered
private if at least 1 parent had any private insurance, regardless of the other parents in- surance
status or type; parental insurance was considered public if both parents had public insurance only, or
1 parent had public only and the other parent was uninsured; parental insurance was considered
uninsured if both par- ents were uninsured. If the parent and child had the same type of health
insurance, their coverage was considered concordant, and if the insurance type was different between
parent and child, their coverage type was considered discordant.
We based household income stratifications on estab- lished MEPS-HC categories. We defined
low income as less than 200% of the federal poverty level (FPL),
combining the MEPS-HC poor, near-poor, and low cate- gories; middle income as 200% to less than
400% FPL; and high income as $400% FPL.15 The FPL for a family of 4 was $16,450 in 1998 and
$22,350 in 2011.20,21
ANALYSIS
All analyses were stratified by family income categories. We do not report results from highincome families because the majority (88%) had private insurance for both child and parent, and
all categories had either no sta- tistically significant changes or too few subjects to assess changes (n
< 30). We used sampling stratification vari- ables, design weights, and a robust variance estimator
in accordance with MEPS guidelines to account for the com- plex sample design of the survey. This
accounts for both the intracluster correlation of children within families and in- traperson correlation
across years.22
We examined the following demographic characteristics for the entire study period as one pooled
sample and report the weighted percentage of each characteristic: age (child categories 04, 59,

1013, 1417 years; parent categories #24, 2544, $45 years), child race/ethnicity (non-Hispanic
white, non-Hispanic nonwhite, Hispanic), region (North, Midwest, South, West), parental education
(<12 years or $12 years), family composition (1 parent or 2 parents), parental employment
(currently employed or unemployed), and childs perceived health status (excel- lent/very good or
good/fair/poor). We conducted descrip- tive analyses of the prevalence of all 9 possible patterns of
coverage type for children paired with a parent or parents, as well as concordant versus
discordant insurance coverage. We assessed differences in the distribution of childparent health
insurance type between 1998 and
2011 with chi-square tests of association using SUDAAN
11.0.1 (Research Triangle Institute, Research Triangle
Park, NC).
We used joinpoint regression (sometimes called piece- wise regression or segmented regression) to
determine if and when coverage patterns showed significant changes throughout the entire study
period (Joinpoint Regression Software 4.0.4, May 2013; Statistical Methodology and Applications
Branch, Surveillance Research Program, National Cancer Institute).23 Joinpoint regression is often
used for 2 simultaneous goals: to identify statistically sig- nificant changes in trend over time (in
direction or rate of decrease or increase) and to quantify that change through an annual percentage
of change statistic. This approach has been used to assess temporal trends in health insurance and
other health care outcomes.2426 The null hypothesis in this analysis was no change in trend, and
the alternative hypothesis was a significant increase or decrease in the prevalence of each health
insurance coverage pattern. The minimum number of joinpoints allowed was 0 (ie, a straight line
over time), indicating no change in child and parent health insurance coverage patterns over time.
The maximum number of joinpoints was set at 2 on the basis of an algorithm taking into account
the number of time points available,23 with one exception: the child

ACADEMIC PEDIATRICS
FAMILY
HEALTH INSURANCE TRENDS, 19982011
3
public/parent private, middle-income group was limited to only 1 joinpoint as a result of small cell
sizes (n < 30) in the years 1998 to 2000 and thus had fewer time points avail- able. A Monte Carlo
permutation method was used to select the model with the best fit, and yearly percentage point
changes were calculated for each segment. Statisti- cally significant changes were those that
increased or decreased over time and were significantly different than an annual percentage change
of 0 (no change over time).
To account for potential differences between child and parent health insurance types in our
analysis of change patterns over time, we used trend segments identified in joinpoint regression
analyses with multinomial logistic regression to allow inclusion of potential confounders. In these
models, the child and parent combined health insur- ance type was the outcome variable (9
categories) and year was the primary independent variable. For the low-income models, all 9
categories were included in the outcome var- iable; however, for the middle-income models, the child
private/parent public and child uninsured/parent public cat- egories were excluded as a result of small
and/or zero cells (n < 30). We adjusted for all demographic characteristics examined, as they are
known to influence health insurance
coverage.7,2729 Marginal effects for year were calculated
for each model and are represented as an adjusted yearly percentage point change. Multinomial
logistic regression models were conducted and marginal effects were calculated using STATA 11.2
IC (StataCorp, College Station, Tex).
Throughout this article, we do not report estimates based on sample sizes of fewer than 30 because
estimates based on such sample sizes are not reliable. A P value of <.05 was considered statistically
significant. Our institutional review board deemed the study exempt.
RESULTS
Children from this sample of low- and middle-income families in the United States predominantly
lived in the South, had excellent or very good health status, were non-Hispanic white, had
parents with $12 years of educa- tion, had employed parents, and had 2 parents living in the
household. Children and parents from low-income families tended to be younger than those from
middle-income fam- ilies (Table 1).
Among children from low- and middle-income families, the prevalence of full-year childparent
health insurance type changed significantly between 1998 and 2011 for several groups including the
following (child type/parent type): private/private; public/private, public/public, pub- lic/uninsured,
uninsured/private, and uninsured/uninsured (Table 2).
TRENDS IN PRIVATE INSURANCE COVERAGE
The percentage of childparent pairs with private insur- ance decreased steadily from 29.2% in
1998 to 19.1% in 2011 for low-income families (<200% FPL), the yearly unadjusted percentage
point decrease was 0.86 per year (95% confidence interval [CI] 1.10, 0.63). The
prevalence of middle-income pairs with child and parent privately insured fell significantly from
74.5% in 1998 to
66.3% in 2011, the yearly unadjusted percentage point decrease was 0.73 (95% CI, 0.98, 0.48).
TRENDS IN UNINSURANCE
The prevalence of low-income childparent pairs with both child and parent uninsured showed
a decrease overall from 26.3% in 1998 to 14.8% in 2011, the yearly unadjusted percentage point
decrease was 0.81 (95% CI, 1.05, 0.58). Middle-income uninsured pairs also saw a decrease
from 13.2% in 1998 to 8.7% in 2011; from 1998 to 2004, there was a yearly unadjusted
percentage point decrease of 0.58 (95% CI, 1.01,
0.15) and from 2007 to 2011, the yearly percentage point decrease was 1.36 (95% CI, 2.19,
0.54).

TRENDS IN DISCORDANT INSURANCE COVERAGE

Overall, discordant coverage increased from 23.2% to
42.1% for low-income families. The prevalence of publicly insured children with uninsured parents
increased from
10.4% in 1998 to 27.2% in 2011, the yearly unadjusted per- centage point increase was 1.74 (95% CI,
1.36, 2.12) from
1998 to 2003, which continued rising, though less dramatically, by 0.98 (95% CI, 0.75, 1.20) percentage points per year between 2003 and 2011. Discordant
coverage increased from 9.3% to 18.8% for middle-income families. The prevalence of publicly
insured children with uninsured parents increased from 1.4% in 1998 to 6.7% in 2011, the yearly
unadjusted percentage point increase was 0.42 (95% CI, 0.34, 0.51).
Yearly percentage point changes from all adjusted multinomial logistic regression models were consistent in magnitude and direction with those found in
joinpoint regression with 2 exceptions: the trends were similar in direction but were no longer
statistically significant for middle-income families with publicly insured children and privately
insured parents, and low-income families with uninsured children and publicly insured parents
(Figs. 1 and 2; Table 3).
DISCUSSION
Type of health insurance coverage patterns changed significantly for low- and middle-income US
children and their parents from 1998 to 2011. Families saw significant decreases in the percentage of
childparent pairs with full- year, private health insurance and pairs without coverage. This coincided
with a significant increase in the percentage of families with discordant coverage. Specifically, we
found an increase in publicly insured children with uninsured par- ents, suggesting that when families
lost private coverage, they were able to obtain public health insurance for their children only.
Decreases in private coverage, concurrent with increases in public coverage could be due to what is
termed crowd outthe movement of privately insured indi- viduals to public insurance.30 Reports of
crowd out have been mixed. For example, one study found that for every
100 children who became eligible for public insurance,

DEVOE ET AL

ACADEMIC PEDIATRICS

Table 1. Demographic Characteristics by Family Income, 19982011*

Low Income (n 65,496)
Income (n 33,246)

Middle

Characteristic
Childs age
Unweighted n
%
04 y
7,831
59 y
9,352
1013 y
7,915
1417 y
8,148
Parents age
#24 y
1,390
2444 y
26,632
$45 y
5,224
Region
Northeast
4,878
Midwest
7,397
South
11,715
West
9,256
Childs health status
Excellent/very good
27,506
Good/fair/poor
5,729
Childs race/ethnicity
Non-Hispanic white
17,005
Non-Hispanic nonwhite
7,278
Hispanic
8,963
Parents education
$12 y
29,929
<12 y
3,242
Family composition

Weighted %

Unweighted n

Weighted

18,895

30.4

19,783

29.7

14,480

21.3

12,338

18.6

7,247

11.7

51,846

78.5

6,403

9.8

8,294

15.1

10,707

20.2

26,327

38.9

20,168

25.9

47,802

75.2

17,677

24.8

16,388

40.4

18,577

27.7

30,531

31.9

41,201

71.0

24,045

29.0

24.9
27.9
23.2
24.0
4.0
79.8
16.2
16.7
24.5
35.5
23.3
84.3
15.7
66.2
17.9
15.9
93.7
6.3

DEVOE ET AL

One parent
7,078
Two parents
26,168
Parents employment
Employed
32,068
Unemployed
1,161

ACADEMIC PEDIATRICS

28,003

42.1

37,493

57.9

49,903

78.3

15,492

21.7

20.7
79.4
97.1
2.9

*Data source: Medical Expenditure Panel SurveyHousehold Component survey.

Weighted percentages are reported for the entire study period (19982011) as 1 pooled
sample. Column percentages are approximately 100% (rounded to nearest tenth of a
percent). Low income, <200% federal poverty level (FPL); middle income, 200% to
<400% FPL. The FPL for a family of four in 2011 was
$22,350.
Sample sizes do not add to total unweighted column number as a result of exclusion of
small number of missing responses.
4 gained public coverage, yet only 2 of them were uninsured before gaining coverage; thus, crowd out
explained half of the increase.31 Other studies, however, found little or no evidence of crowd
out.32,33 In this study, we saw a much larger increase in publicly insured children with uninsured
parents compared to publicly insured children with privately insured parents. Thus, the increasing
cost of private health insurance, coupled with reductions in employer-sponsored insurance offerings
and the historical lack of opportunities for adults to gain public coverage, likely account for the
changes reported here.
When faced with unaffordable coverage options and/or reductions in benefit packages, low- and
middle-income families were forced to look beyond employer-sponsored, private coverage.34,35
During the time period studied, many states expanded their childrens health insurance programs
while simultaneously limiting eligibility and public insurance enrollment opportunities for
adults.36
The changes in health insurance type seen in this study were not explained by differences in child or
parent charac- teristics, as evidenced by the consistent results between un- adjusted and adjusted
analyses. Further research is needed
to assess the impact of these increasing discordant family coverage patterns.
IMPLICATIONS
Despite improvements in childrens coverage rates, this study suggests a trend of coverage loss for
parents that could negatively affect the whole family. Insured children with uninsured parents have
higher odds of experiencing health insurance coverage gaps and unmet health care needs compared to
insured children with insured parents.8,37 The Affordable Care Act (ACA) calls for state Medicaid
program expansions to cover adults earning #138% of the FPL and for health insurance marketplaces
to allow individuals not offered health insurance through their employer the ability to purchase
coverage on their own; millions have gained coverage through these new opportunities.38,39 Few
coverage options are available for low-income adults living in states that chose not to expand their
Medicaid programs.40 Without expansions in Medicaid, uninsured parents will need to rely on
private health insurance through health insurance marketplaces. Though federal tax credits exist
for low- and middle-

ACADEMIC PEDIATRICS

FAMILY HEALTH INSURANCE TRENDS, 19982011

Table 2. Percentage of Child and Parent Full-Year Health Insurance Type by Family Income,
1998 Versus 2011
Low Income (n 65,496)
Income (n 33,246)
Full-Year Health
Child
type/parent type
Private/private
Private/public
Private/uninsured
Public/private
Public/public
Public/uninsured
Uninsured/private
Uninsured/public
Uninsured/uninsured
Child/parent insurance
concordancek
Concordant
Discordant

1998 Weighted
29.
2*
.

2.
0
3.4
*
21.
3
10.
4*
4.7
2.*
1
26.
3*

2011 Weighted
19.
1*
.

2.
4
8.2
*
24.
0
27.
2*
2.5
1.*
3
14.
8*

Middle
1998 Weighted
74.
5*
.

2.
4
2.9

3.0
*
1.4
4.*
3
.

13.
2*

2011
66.3*
.

1.
9
4.
3
6.2*
6.7*
5.
4
.

8.7*

76.
57.
90.
81.2*
23.
42.
9.3
18.8*
2*
1*
*
*P value (P < .05) considered statistically significant, calculated by chi-square tests,
comparing rates in 1998 vs 2011.
Data source: Medical Expenditure Panel SurveyHousehold Component survey. Column
percentages are approximately 100% (rounded to nearest tenth of a percent). Low income,
<200% federal poverty level (FPL); middle income, 200% to <400% FPL. The FPL for a family
of four in 2011 was $22,350.
Estimates not reported due to small cell sizes (n < 30).
Cell sizes for years 1998 to 2000 were <30; this value is from the year 2001.
kConcordant indicates child private/parent private; child public/parent public; child
uninsured/parent uninsured; discordant, child private/ parent public; child private/parent
uninsured; child public/parent private; child public/parent uninsured; child uninsured/parent
private; child uninsured/parent public.
income families to help pay for marketplace premiums, cost is still reported to be a barrier to
coverage.41
The unknown future of CHIP is cause for concern for low- and middle- income families.42
Without CHIP, millions of children may become uninsured through the family glitch (ie, adults
would not quality for ACA sub- sidies because they have the income to afford coverage for
themselves, even if they cannot afford the cost of
the premium for the family).43,44 This study uncovered a disturbing historical trend in families
insurance coverage: as children gained coverage, parents lost coverage at an alarming rate. Thus, as
changes in health insurance options and eligibility continue to occur, it will remain important to
monitor the stability of family coverage. In addition to demonstrating novel methods for this
continued evaluation of family coverage patterns, we

ACADEMIC PEDIATRICS

FAMILY HEALTH INSURANCE TRENDS, 19982011

Figure 1. Trends in child and parent full-year health insurance type, low income. Symbols
represent actual percentage, while lines represent the trend per joinpoint regression. Change
in trend is identified through joinpoint regression (P < .05). Low income, <200% federal
poverty level (FPL). FPL in 2011 was $22,350. Source: Medical Expenditure Panel Survey,
19982011.

DEVOE ET AL

ACADEMIC PEDIATRICS

Figure 2. Trends in child and parent full-year health insurance type, middle income. Symbols
represent actual percentage, while lines repre- sent the trend per joinpoint regression.
Change in trend is identified through joinpoint regression (P < .05). Middle income, 200% to
400% federal poverty level (FPL). FPL in 2011 was $22,350. Source: Medical Expenditure
Panel Survey, 19982011.
Table 3. Unadjusted and Adjusted Yearly Percentage Point Change in Child and Parent
Health Insurance Type by Family Income,
19982011
Full-Year Health Insurance Type,
Child Type/Parent Type
Low income

Joinpoint Identified Trend

Yearly Percentage Point Change (95% CI) Unadjusted

Private/priv
ate
Private/pub
Public/public
Public/uninsured
Uninsured/private
Uninsured/public
Uninsured/uninsured
Private/private
Private/public
Private/uninsured
0.07,
0.02)
Public/private
Public/public
Public/uninsured

1998
2011
.
20042009
20092011
19982011
19982003
20032011
19982011
19982011
19982011
1998
2011
0.02 ( 0.08,
0.03)
20012011
19982011
19982011

0.86 ( 1.10,
0.63)*
.
0.36 ( 0.97,
0.24)
0.94 ( 0.81,
2.68)
1.74
0.43 (1.36,
(0.26, 2.12)*
0.59)*
0.98 (0.75, 1.20)*
0.15 ( 0.22,
0.07)*
0.04 ( 0.08,
0.01)*
0.73 ( 0.98,
0.48)*
19982011
0.14 (0.03, 0.24)*
0.16
0.42 (0.06,
(0.34, 0.25)*
0.51)*

Adjusted
0.92 ( 1.12,
0.72)*
.
0.35 ( 0.72,
0.02)
0.86 ( 0.04,
1.75)
1.63
0.53 (1.18,
(0.27,
2.08)*(0.65,
1.00
1.34)*
0.17 ( 0.25,
0.09)*
0.03 ( 0.08,
0.01)
0.61 ( 0.83,
0.40)*
0.02 (
0.11 ( 0.03,
0.25)
0.37
0.22 (0.28,
(0.10,
0.47)*

DEVOE ET AL

Uninsured/private
Uninsured/public
Uninsured/uninsured

1998
2011
.
k
1998

0.02 ( 0.06,
0.11)
.
k
0.58 ( 1.01,

ACADEMIC PEDIATRICS

0.01 ( 0.10,
0.11)
.
k
0.66 ( 1.12,

CI indicates confidence interval.

*P value (P < .05) considered statistically significant.
Data source: Medical Expenditure Panel SurveyHousehold Component survey. Low
income, <200% federal poverty level (FPL); middle income, 200% to <400% FPL. The FPL in
2011 was $22,350.
Unadjusted results from joinpoint regression.
Adjusted results from multinomial logistic regression, with covariates including childs age,
parent age, race/ethnicity of the child, region of residence, parental education, family
composition, parental employment, and childs perceived health status.
kSmall sample sizes (n < 30) for most years; estimates are unreliable.

ACADEMIC PEDIATRICS

FAMILY HEALTH INSURANCE TRENDS, 19982011

also demonstrate how joinpoint analyses can be used in future analyses for researchers to track
longitudinal changes in the slope and direction of trends in health insurance coverage.
LIMITATIONS
Our analyses were limited by existing MEPS-HC vari- ables. As with all self-reported data,
response bias remains a possibility. However, the MEPS-HC asks several ques- tions about health
insurance status and type at various time points, and survey staff logically edit responses for
consistency across variables. The MEPS is a nationally representative data set that does not
account for state- level differences stemming from individual state policies, which differentially
expanded and contracted public health insurance programs during the study time period.
CONCLUSIONS
From 1998 to 2011, low- and middle-income US fam- ilies experienced a decrease in the
percentage of child parent pairs with private health insurance and pairs without insurance.
Concurrently, there was a rise in discordant coverage patterns, mainly publicly insured children with
uninsured parents.
ACKNOWLEDGMENTS
This study was financially supported by the Agency for Healthcare Research and Quality (AHRQ)
(grant 1 R01 HS018569), Patient-Centered Outcomes Research Institute (Health Systems Cycle I,
2012), the National Cancer Institute of the National Institutes of Health (grant 1 R01
CA181452 01), the Oregon Health & Science University Department of
Family Medicine, and the Ohio State University Department of Family Medicine. The funding
agencies had no involvement in the design and conduct of the study; analysis, and interpretation of
the data; and prepara- tion, review, or approval of the manuscript. AHRQ collects and manages the
MEPS.
REFERENCES
1. Sommers BD, Baicker K, Epstein AM. Mortality and access to care among adults after state
Medicaid expansions. N Engl J Med. 2012;
367:10251034.
2. Asplin BR, Rhodes KV, Levy H, et al. Insurance status and access to urgent ambulatory care
follow-up appointments. JAMA. 2005;294:
12481254.
3. Burstin HR, Lipsitz SR, Brennan TA. Socioeconomic status and risk for substandard medical
care. JAMA. 1992;268:23832387.
4. Bindman AB, Grumbach K, Osmond D, et al. Preventable hospitalizations and access to health care. JAMA. 1995;274:305311.
5. Vistnes J, Schone B. Pathways to coverage: the changing roles of pub- lic and private sources.
Health Aff. 2008;27:4457.
6. Kaiser Commission on Medicaid and the Uninsured. The Uninsured
A Primer: Key Facts About Health Insurance on the Eve of Health Re- form. Menlo Park, Calif:
Henry J. Kaiser Family Foundation; 2013.
7. DeVoe J, Krois L, Edlund C, et al. Uninsurance among children whose parents are losing
Medicaid coverage: results from a statewide survey of Oregon families. Health Serv Res.
2008;43(1 Pt 2):401418.
8. DeVoe JE, Tillotson CJ, Wallace LS. Childrens receipt of health care services and family health
insurance patterns. Ann Fam Med. 2009;7:
406413.

ACADEMIC PEDIATRICS

FAMILY HEALTH INSURANCE TRENDS, 19982011

9. DeVoe JE, Tillotson CJ, Angier H, et al. Predictors of childrens health insurance coverage
discontinuity in 1998 versus 2009: parental coverage continuity plays a major role. Matern Child
Health J. 2015;
19:889896.
10. DeVoe JE, Marino M, Angier H, et al. Effect of expanding Medicaid for parents on childrens
health insurance coverage: lessons from the Oregon experiment. JAMA Pediatr.
2015;169:e143145.
11. Cohen RA, Makuc DM, Bernstein AB, et al. Health insurance
coverage trends, 19592007: estimates from the National Health
Interview Survey. Natl Health Stat Report. 2009;17:125.
12. DeVoe JE, Crawford C, Angier H, et al. The association between pub- lic coverage for children
and parents persists, 20022010. Matern Child Health J. 2015;19:17661774.
13. Ziegenfuss JY, Davern ME. Twenty years of coverage: an enhanced
current population survey19892008. Health Serv Res. 2011;46(1
Pt 1):199209.
14. Berdahl TA, Friedman BS, McCormick MC, et al. Annual report on health care for children and
youth in the United States: trends in racial/ethnic, income, and insurance disparities over time,
2002
2009. Acad Pediatr. 2013;13:191203.
15. Agency for Healthcare Research and Quality. Medical Expenditure Panel Survey: household
component. Available at: http://meps.ahrq. gov/mepsweb/survey_comp/household.jsp; 2010.
Accessed February
27, 2014.
16. Cohen J, Monheit A, Beauregard K, et al. The Medical Expenditure
Panel Survey: a national health information resource. Inquiry.
1996-1997;33:373389.
17. Lau JS, Adams SH, Park MJ, et al. Improvement in preventive care of young adults after the
Affordable Care Act: the Affordable Care Act is helping. JAMA Pediatr. 2014;168:11011106.
18. Gold R, DeVoe JE, McIntire PJ, et al. Receipt of diabetes preventive
care among safety net patients associated with differing levels of insurance coverage. J Am
Board Fam Med. 2012;25:4249.
19. Olson LM, Tang SF, Newacheck PW. Children in the United States
with discontinuous health insurance coverage. N Engl J Med. 2005;
353:382391.
20. US Department of Health and Human Services. Annual update of the
HHS poverty guidelines. Fed Reg. 1998;63:92359238.
21. US Department of Health and Human Services. Annual update of the
HHS poverty guidelines. Fed Reg. 2011;76:36373638.
22. Williams RL. A note on robust variance estimation for cluster- correlated data. Biometrics.
2000;56:645646.
23. Kim HJ, Fay MP, Feuer EJ, et al. Permutation tests for joinpoint
regression with applications to cancer rates: correction. Stat Med.
2000;19:335351.
24. Wong JR, Harris JK, Rodriguez-Galindo C, et al. Incidence of child- hood and adolescent
melanoma in the United States, 19732009. Pediatrics. 2013;131:846854.
25. Simon AE, Uddin SG. National trends in primary cesarean delivery,
labor attempts, and labor success, 19902010. Am J Obstet Gynecol.
2013;209:554.e551554.e558.
26. Barnes RFW, Moore ML, Garfein RS, et al. Trends in mortality of tuberculosis patients in the
United States: the long-term perspective. Ann Epidemiol. 2011;21:791795.

ACADEMIC PEDIATRICS

FAMILY HEALTH INSURANCE TRENDS, 19982011

27. Hill HD, Shaefer HL. Covered today, sick tomorrow? Trends and correlates of childrens health insurance instability. Med Care Res Rev.
2011;68:523536.
28. Sommers BD. Insuring children or insuring families: do parental and sibling coverage lead to
improved retention of children in Medicaid and CHIP? J Health Econ. 2006;25:11541169.
29. Yamauchi M, Carlson MJ, Wright BJ, et al. Does health insurance
continuity among low-income adults impact their childrens insur- ance coverage? Matern Child
Health J. 2013;17:248255.
30. Cutler DM, Gruber J. Does public insurance crowd out private insurance? Q J Econ. 1996;111:391430.
31. Gresenz CR, Edgington SE, Laugesen M, et al. Take-up of public in- surance and crowd-out of
private insurance under recent CHIP expan- sions to higher income children. Health Serv
Res. 2012;47:
19992011.
32. Hamersma S, Kim M. Participation and crowd out: assessing the ef- fects of parental Medicaid
expansions. J Health Econ. 2013;32:
160171.

DEVOE ET AL

ACADEMIC PEDIATRICS

33. McMorrow S, Kenney GM, Waidmann T, et al. Access to private coverage for children enrolled
in CHIP. Acad Pediatr. 2015;15(3 suppl):S50S55.
34. Young RA, DeVoe JE. Who will have health insurance in the future?
An updated projection. Ann Fam Med. 2012;10:156162.
35. Gould E. Employer-sponsored health insurance erosion continues in
2008 and is expected to worsen. Int J Health Serv. 2010;40:743776.
36. Kaiser Commission on Medicaid and the Uninsured. States respond to fiscal pressure: state
Medicaid spending growth and cost containment in fiscal years 2003 and 2004. Results from a
50-state survey. Available at: http://www.kff.org/medicaid/upload/States-Respond-to- FiscalPressure-State-Medicaid-Spending-Growth-and-Cost-Contain ment.pdf; 2003. Accessed August 3,
2011.
37. DeVoe JE, Krois L, Edlund C, et al. Uninsured but eligible children:
are their parents insured? Recent findings from Oregon. Med Care.
2008;46:38.
38. Henry J, Kaiser Family Foundation. Recent Trends in Medicaid and CHIP Enrollment as of
January 2015: Early Findings from the CMS Performance Indicator Project. Menlo Park,
Calif: Henry J. Kaiser Family Foundation; 2015.
39. Henry J, Kaiser Family Foundation. Data Note: How Has the Individ- ual Insurance Market
Grown Under the Affordable Care Act?. Menlo Park, Calif: Henry J. Kaiser Family Foundation;
2015.
40. Henry J, Kaiser Family Foundation. Implementing the ACAs Medicaid-Related Health
Reform Provisions After the Supreme Courts Decision. Menlo Park, Calif: Henry J. Kaiser
Family Founda- tion; 2012.
41. Henry J, Kaiser Family Foundation. Adults Who Remained Uninsured at the End of 2014. Menlo
Park, Calif: Henry J. Kaiser Family Foun- dation; 2015.
42. Kaiser Family Foundation. Childrens health coverage: Medicaid,
CHIP and the ACA. Available at: http://kff.org/health-reform/issue- brief/childrens-healthcoverage-medicaid-chip-and-the-aca/; 2014. Accessed February 23, 2015.
43. Health Policy Brief: the family glitch. Health Affairs, November 10,
2014, Available at: http://www.healthaffairs.org/healthpolicybriefs/
brief.php?brief_id129. Accessed June 18, 2015.
44. Selden TM, Dubay L, Miller GE, et al. Many families may face sharply higher costs if public
health insurance for their children is rolled back. Health Aff. 2015;34:697706.