Increased Rates of Morbidity, Mortality, and Charges for Hospitalized Children With Public or No Health Insurance as Compared With Children With Private Insurance in Colorado and the United States
BACKGROUND. There has been a gradual decrease in the proportion of children covered by private health insurance in Colorado and the United States with a commensurate increase in those with public insurance or having no insurance which may impact access to care and outcomes.
OBJECTIVE. The purpose of this work was to determine whether children with public or no health insurance have differences in hospital admission rates, morbidity, mortality, and/or charges that might be improved if standards of primary care comparable to those of children with private insurance could be achieved.
METHODS. We conducted a retrospective comparison of hospitalization-related outcomes for children <18 years of age in Colorado from 1995–2003 and in the United States in 2000. Population-based rates for hospital admission were determined stratified by age, race/ethnicity, disease grouping, and health insurance status.
RESULTS. Compared with those with private insurance, children in Colorado and the United States with public or no insurance have significantly higher rates of total hospital admission, as well as admission for chronic illness, asthma, diabetes, vaccine-preventable disease, psychiatric disease, and ruptured appendix. These children have higher mortality rates, higher severity of illness, are more likely to be admitted through the emergency department and have significantly higher hospital charges per insured child. Higher hospitalization rates occur in children who are nonwhite and/or Hispanic and those who are younger. If children with public or no health insurance in the United States in 2000 had the same hospitalization outcomes as children with private insurance, $5.3 billion in hospital charges could have been saved.
CONCLUSIONS. There is an opportunity to achieve improved health outcomes and decreased hospitalization costs for children with public or no health insurance if private insurance standards of health care could be achieved for all US children.
The US Census reports that 34.1% of children <18 years of age were not covered by private health insurance at some time during 2003, a percentage that has been increasing in recent years; one third of these children went without any health insurance, with the remainder having some coverage through public health insurance programs, such as Medicaid and/or State Children's Health Insurance Program (SCHIP).1 It has been estimated that an additional 7.7% of children with some health insurance coverage during the year may suffer from temporary gaps in that coverage.2 It is reasonable to assume that lack of health insurance or gaps or insufficiency in its coverage may result in inferior health outcomes for children. Although it is widely believed that per capita health care expenditures are higher for children who are uninsured or are enrolled in public insurance programs compared with those children with private commercial health insurance, available data quantifying this difference are limited. We evaluated this hypothesis by analyzing hospitalization rates for children with private insurance as compared with those with public insurance or no insurance for a range of diseases and conditions to determine whether children with public or no health insurance have differences in hospital admission rates, morbidity, mortality, and/or charges that might be improved if standards of care comparable to those of children with private insurance could be achieved.
This was a population-based analysis of hospitalization rates for children who were >28 days and <18 years of age in Colorado from 1995–2003 and separately for children in the United States in 2000. These 2 different populations were selected because Colorado data permitted comparison of the consistency of outcome measures over an extended period of time, whereas data accessibility limited extrapolation to the entire United States for only a single year. Hospitalization rates in Colorado were calculated using numerator data obtained from the Colorado Hospital Association (CHA) state inpatient database that contains standard administrative hospital discharge data grouped using the all-patient refined diagnosis-related group (APR-DRG) version 15 grouper codeveloped by the National Association of Children's Hospitals and Related Institutions and 3M Health Information Systems (Salt Lake City, UT). Diagnostic groupings were created using major diagnostic categories (MDCs) and International Classification of Diseases, Ninth Revision (ICD-9) diagnostic codes based on published precedents when possible (see “Appendix”).3,4 Hospitalizations for MDC 14 (pregnancy-related conditions) and MDC 15 (birth-related conditions) were excluded. Health insurance status for each hospitalized child was grouped as private (Blue Cross/Blue Shield, commercial insurance/indemnity plans/self insured, other liability insurance/no fault/casualty, or health maintenance organization-preferred provider organization/managed care/discounted) or public/none (self-pay, no charge/charity/research, other, Colorado Medically Indigent, Medicare, Medicaid, worker's compensation, Champus, or other government).
A preliminary analysis of Medicaid/SCHIP hospitalization rates tended to overestimate the difference between public insurance outcomes as compared with private insurance, because it included a disproportionate number of the hospitalizations of initially uninsured children who were retroactively qualified for Medicaid without a similar adjustment being possible for the denominator (at the major pediatric Medicaid hospital provider in Colorado, reclassification to Medicaid status is estimated to occur in >50% of initially “uninsured” cases). We, therefore, elected to combine uninsured hospitalized children with those having public insurance to better reconcile available data for rate calculations. No adjustment for multiple hospital admissions for individual children could be made because of the lack of identifiers within the CHA database.
Using DataFerrett, a collaborative tool developed by the US Bureau of the Census and the Centers for Disease Control to extract census data to estimate health insurance coverage, we obtained estimates for Colorado children <18 years of age from 1995–2003 grouped as private insurance or public/none (calculated as total children <18 years minus those with private insurance during any portion of that year) commensurate with the numerator classification combining uninsured and publicly insured children.5 Denominator estimates included children <28 days of age. We were able to obtain these estimates for insured populations in the United States broken down by age category and race/ethnicity; however, reliable denominator stratification at the state level was not possible because of the large confidence intervals associated with decreasing sample size.
US hospitalization rates for children <18 years of age were calculated by using identical methods and definitions using weighted hospitalization estimates from the Kids' Inpatient Database (KID) from the Healthcare Cost and Utilization Project for the year 2000.6 Age stratification of results was based on categories determined by the KID 2000 data structure, because a number of states provided only age-aggregated data to the KID 2000 database. Stratification by race and ethnicity was performed using consistent coding used by the CHA and KID. Two groups were created: white/non-Hispanic and nonwhite and/or Hispanic, the latter group including all of those indicated as Hispanic, Indian, Asian, and/or black.
Monthly insurance coverage denominator data were available from the Colorado Department of Health Care Policy and Financing for Colorado children covered by Medicaid from July 1997 through November 2003 that provided estimates for the numbers of children who had selected or were assigned to a primary care physician (health maintenance organization or primary care physician program) as compared with those in an unassigned fee-for-service (UFFS) Medicaid category. This permitted a correlation of Medicaid hospital admissions in the CHA database with the number and percentage of children with each type of Medicaid insurance by month over this time frame.
Hospitalization rates were calculated for: all hospitalizations; hospitalization via the emergency department; children hospitalized with an APR-DRG severity score >2; chronic disease; and ambulatory care-sensitive conditions, including asthma, diabetes, vaccine preventable disease (excluding influenza), psychiatric disease (MDC 19), and appendicitis because of a ruptured appendix or with peritonitis (see “Appendix”).3 For Colorado, total hospital charges for years 1995 through 2002 were adjusted to 2003 prices using the Consumer Price Index. US and regional hospitalization charges were based on the actual charge estimate for the year 2000 as recorded in the KID 2000 database. Hospital charges per insured child were calculated by dividing the total hospital charges for hospitalized children by the total number of children in each insured group. The difference between the average hospital charges per child in the public or no-insurance group compared with the private insurance group multiplied by the total number of covered individuals in the public or no-insurance group yielded the potential excess hospital charge gap between these 2 populations. This gap represents the maximum savings that could potentially be achieved for children with public or no insurance if they had improved outcomes equivalent to children with private insurance. It was assumed that reported hospital charges represented some multiple of actual hospital costs.
The numerator data for Colorado rate comparisons represented the actual numbers of reported events. Population estimates for health insurance coverage had relatively small SE estimates. We, therefore, compared the differences between the hospitalization rates for each year for children with private insurance and those with public or no insurance over the years 1995–2003 using the paired-samples t test for comparison of mean differences of SPSS 13 (SPSS Inc, Chicago, IL). For each disease category we defined “rate ratio” as the mean yearly rate for those with public or no health insurance for the years 1995–2003 divided by the rate of the analogous years for those with private insurance. Comparisons for US and regional rates for 2000 were calculated using the χ2 test (Statcalc in EpiInfo 6, Centers for Disease Control and Prevention, Atlanta, GA) and reported as relative risks.
Logistic regression analysis for hospitalized Colorado children was conducted with insurance status as the dependent variable and race/ethnicity and chronic disease as the independent variables using SPSS 13. Linear regression was similarly accomplished for the monthly number of hospitalized Medicaid children correlated with the total number of children covered by Medicaid and the proportion of children within the UFFS group.
Table 1 shows the demographic characteristics of the populations of hospitalized children in Colorado for 2003 (the last year of our analysis) and the United States for 2000. The age distribution of hospitalized children in Colorado and the United States with public or no health insurance was significantly younger (P < .0001) than those with private health insurance with the former weighted toward young children and the latter toward older adolescents. There was a marked difference in race/ethnicity distribution of children with public or no health insurance in both Colorado and the United States with approximately two thirds of hospitalized children falling into the nonwhite and/or Hispanic category as compared with one third for those with private health insurance (P < .0001).
Colorado Hospitalization Rates
Figure 1 demonstrates a relatively stable rate of overall hospital admission for Colorado children with private insurance from 1995–2003 with consistently higher rates for children with public or no health insurance. Table 2 validates this observation, showing significantly higher hospitalization rates per 100000 insured for children with public or no health insurance for: all hospitalizations from 1995–2003 (mean rate difference: 1333; 95% confidence interval [CI]: 873 to 1792) hospitalizations via the emergency department, higher severity hospitalizations (ie, APR-DRG severity score >2), chronic diseases, asthma, diabetes, psychiatric disease, and vaccine preventable disease, with the comparative rate ratio ranging between 1.46 and 2.17 times the private insurance rates. Comparisons of the difference between hospitalization rates stratified by APR-DRG MDC category from 1995–2003 (data not shown) for privately insured children and those with public or no insurance did not demonstrate significantly increased rates for MDCs 8, 16, 17, or 25 (musculoskeletal system and connective tissue; blood, blood forming organs, immunologic disorders; myeloproliferative, poorly differentiated neoplasms; or multiple significant trauma).
The percentage of children hospitalized via the emergency department was also significantly higher in children with public or no health insurance, whereas the hospitalization rate for appendectomy (rate ratio: 1.07; 95% CI: −3.7 to 16.8) was not significantly different between the 2 categories. The percentage of children with appendicitis who had a ruptured appendix and/or peritonitis was significantly higher in children with public or no health insurance. Children with public or no health insurance also had a significantly higher mortality rate per 100000 insured children (mean rate difference: 6.5; 95% CI: 4.1 to 8.8; P < .001). Stepwise logistic regression analysis showed significant (P < .002) independent contributions from nonwhite and/or Hispanic race/ethnicity (β = .812), admission year (β = −.010), age in years (β = −.047), chronic disease (β = .042), and ambulatory-sensitive conditions (β = .202) for hospitalization of children with public or no health insurance. Chronic disease comprised 11.9% of all hospital admissions for children with public or no health insurance.
For the period from July 1997 through November 2003, when a breakdown of types of Medicaid coverage for Colorado children was available (primary care physician program, managed care, and UFFS), stepwise linear regression showed a significant direct correlation with monthly Medicaid admission rates and both the total number of children with Medicaid insurance (standardized β = .248; P = .028) and the proportion of those with UFFS classification (standardized β = .292; P = .009).
Colorado Hospital Charges
There were considerably higher CPI-adjusted hospital charges per insured child for the public or no insurance group compared with the private insurance group every year of the analysis (Table 2). The mean difference in excess adjusted charges was $128 per insured child for children with public or no health insurance as compared with those with private insurance over that time period. This differential in hospital charges per insured child with public or no insurance represented an estimated excess expenditure of $46.1 million in 2003 for the state of Colorado as compared with those with private insurance.
US and Regional Hospitalization Rates
Table 3 shows similarly increased hospitalization rates for children with public or no health insurance for the United States for 2000 as compared with privately insured children in disease categories comparable to the longitudinal Colorado state analysis. Overall, there was an increased relative risk of higher mortality (2.38; 95% CI: 2.28 to 2.49) as well as condition-specific hospitalization rates for children with public or no health insurance with relative risks ranging from 1.20 to 2.37 times more than children with private insurance. Nationally there was an estimated excess of 2378 deaths in the US in 2000 among children who had public or no health insurance as compared with those with private health insurance. In this year, there was also an estimated difference in national hospitalization charges for children with public or no health insurance of $242 as compared with children with private health insurance. For the year 2000, this amounted to an estimated $5.3 billion dollars in excess hospitalization charges for children with public or no health insurance as compared with children with private health insurance.
Although regional comparisons consistently showed similar differences for population-based hospitalization morbidity, mortality, and hospital charges between children with private insurance and public or no insurance (data not shown), quantitative differences between regions were also apparent. For example, per capita hospital charges for both groups were lower in the South and Midwest compared with higher per capita charges in the Northeast and West.
Using age categories available in KID 2000, hospitalization rates were calculated for insurance groups as determined by race/ethnicity. Rates were highest in young children (<5 years of age), decreased in school-aged children, and increased in older teenagers (Fig 2). Over the 5 age groups, children with public or no health insurance had significantly higher hospitalization rates in both the nonwhite and/or Hispanic and white/non-Hispanic racial/ethnic groups when compared with age group-aggregated children with private insurance (P = .043, Wilcoxon signed-ranks test). Similarly, over the 5 age groups, nonwhite and/or Hispanic children had significantly higher hospitalization rates in both the public or no health insurance and private insurance groups when compared with age group-matched white/non-Hispanic children (P = .043, Wilcoxon signed-ranks test).
This study documents significantly higher rates of condition-specific, hospitalization-associated morbidity, mortality, and charges for children with public or no health insurance as compared with children with private health insurance in both Colorado and the United States. It is reasonable to hypothesize that inadequate insurance coverage may be one factor that adversely impacts access to preventive and acute primary care resulting in greater emergency department use and hospitalization-associated morbidity and mortality along with their resultant increased costs. In Colorado, physicians have been reimbursed by Medicaid at ∼50% of overhead cost and hospitals at 70% of actual cost.7 Such low public insurance payment rates may result in a reduction in access to primary health care physicians and/or decreased financial viability of providers (physician practices and hospitals) resulting in cost shifting with a commensurate increase in private sector health costs.8–12 Low reimbursement rates and administrative inefficiencies may compromise the ability of Medicaid to assure the equal access statute that mandates that enrollees receive care comparable to that received by children with private insurance.7,13 Our study demonstrates such inequities in hospitalization-associated outcomes.
In a previous study, we demonstrated that children enrolled in the Colorado unassigned fee-for-service Medicaid program were less likely to have a usual source of primary care resulting in decreased primary care visits, decreased preventive care, and poorer vaccination status.7 In this current study, children covered by Medicaid who did not have an assigned or selected primary care physician (UFFS) were associated with a significantly higher hospital admission rate. Similarly, Mitchell and Gaskin14 reported access problems in children with special health care needs with fee-for-service Medicaid coverage as compared with those with managed care coverage. Olson et al2 have shown that children from low-income families are more likely than children from middle- and high-income families to be uninsured or covered by public insurance and less likely to have had a medical office visit or a dental visit, less likely to have medicines prescribed, and more likely to have used an emergency department. Insurance status may also impact access to urgent ambulatory care follow-up appointments.15
Our results show significantly higher hospitalization rates for children with public or no insurance having ambulatory care-sensitive conditions, such as asthma, vaccine-preventable disease, psychiatric conditions, and diabetic ketoacidosis. The fact that insurance-associated differences were identified for these conditions but not those unlikely to be altered by primary care (eg, appendectomy, childhood cancer, trauma, and orthopedic disorders) in Colorado suggests that these results are not artifacts of our rate calculation methodology. This is further supported by the significantly higher percentage of children with public or no insurance admitted through the emergency department and the observation that the rate ratio for children with public or no health insurance hospitalized in Colorado with a ruptured appendix as compared with those with private insurance was 1.25 (P < .001), whereas the difference in appendectomy rates between the insurance groups was minimal (1.07; P = .180). Similar findings have been reported in other studies.3,16–18 O'Toole et al19 reported higher ruptured appendix rates associated with a longer duration of symptoms in children with Medicaid insurance.
Outcome equity is not solely dependent on the quantity or quality of health insurance given the demographic and cultural differences noted between our insured populations. In hospitalized Colorado children, age, race/ethnicity, and chronic disease status all correlated independently with insurance status. Public health insurance programs have a disproportionate number of young children, are disproportionately nonwhite and/or Hispanic, and serve families with low income that qualifies them for Medicaid or SCHIP.20,21 In the United States, our data show that nonwhite and/or Hispanic children have higher hospitalization rates even if they have private insurance. Genetic factors, socioeconomic factors, and environmental conditions may affect the disease burden of some clinical conditions and may also affect care-seeking behaviors and compliance.22,23 Nonetheless, when controlling for age and race/ethnicity, hospitalization rates were still significantly higher in the United States in the public or no insurance group suggesting structural failures in the available system of care.
The observation that children with public or no health insurance have inferior health outcomes when compared with children with private health insurance should not lead to the conclusion that such outcomes could not be improved or that public insurance systems are inherently flawed. Medicaid and SCHIP are methods for paying for health care, not for providing it. We combined the public and no-insurance groups solely because data limitations would have otherwise resulted in an overestimate of the excess morbidity and charges associated with publicly insured children. From a purely financial point of view, if the provision of consistent primary care for all children (including systems to address racial/ethnic differences in the use and efficacy of primary care) could provide the same hospitalization outcomes for the public or no-health insurance children that are achieved for those with private insurance, a substantial proportion of $46 million of excess hospital charges in Colorado in 2003 and $5.3 billion in the United States in 2000 might have been saved and, along with commensurate reductions in emergency department use, used to partially offset the increase in primary care access and services. Cohen12 has shown that having a private-practice physician as a usual source of care can decrease total physician Medicaid expenses by as much as one third. A recent study also shows a marked reduction in emergency department use (and costs) by Medicaid patients with regular access to pediatric care.24 There is evidence that enrollment in SCHIP can improve such access.25,26
The President's “New Freedom” initiative has articulated the essential goals for children with special health care needs that include having access to ongoing, comprehensive health care and adequate insurance for these services,27 and, yet, a recent study demonstrates that >40% of even these most vulnerable children do not meet these criteria.28 Provision of public insurance coverage for children in the United States is not sufficient to assure high-quality outcomes if that coverage has gaps or does not provide access to effective primary and preventive care.29–31 Allocating the dollars associated with excess hospital use of children with public or no health insurance ($5.3 billion in excess charges for the United States in 2000) to improve access to and efficacy of primary care for these same children would not entirely offset the expenditure required for effective health insurance coverage, but it would provide a significant financial incentive for making quality health care for all US children the ethical priority that it should be.
This work was supported by Amer Fund for Community Pediatrics, Children's Hospital (Denver, CO).
We gratefully acknowledge the intellectual contributions of Bernard Nelson, Jessica Bondy, Dennis Lezotte, and Marian Sills in the design and analysis of this study.
- Accepted March 2, 2006.
- Address correspondence to James Todd, MD, Children's Hospital, 1056 E 19th Ave, Denver, CO 80218. E-mail:
This work was presented in part at the annual meeting of the Pediatric Academic Societies; May 15, 2005; Boston, MA.
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵US Bureau of the Census. Income, Poverty, and Health Insurance Coverage in the United States: 2003. Washington, DC: US Bureau of the Census; 2004
- ↵Bratton SL, Haberkern CM, Waldhausen JH. Acute appendicitis risks of complications: age and Medicaid insurance. Pediatrics.2000;106 :75– 78
- ↵US Bureau of the Census. DataFerrett. Washington, DC: US Bureau of the Census; 2005
- ↵Healthcare Cost and Utilization Project. Kids' Inpatient Database. Rockville, MD: Healthcare Cost and Utilization Project; 2004
- ↵Berman S, Armon C, Todd J. Impact of a decline in Colorado Medicaid managed care enrollment on access and quality of preventive primary care services. Pediatrics.2005;116 :1545– 1546
- Morrisey MA. Hospital pricing: cost shifting and competition. EBRI Issue Brief.1993;137 :1– 17
- ↵Cohen JW, Cunningham PJ. Medicaid physician fee levels and children's access to care. Health Aff (Millwood).1995;14 :255– 262
- ↵Korioth T. Medicaid ruling favors “equal access” for Okla. kids. AAP News.2005;26 :1
- ↵Mitchell JM, Gaskin DJ. Do children receiving Supplemental Security Income who are enrolled in Medicaid fare better under a fee-for-service or comprehensive capitation model? Pediatrics.2004;114 :196– 204
- Smink DS, Fishman SJ, Kleinman K, Finkelstein JA. Effects of race, insurance status, and hospital volume on perforated appendicitis in children. Pediatrics.2005;115 :920– 925
- ↵Smink DS, Finkelstein JA, Kleinman K, Fishman SJ. The effect of hospital volume of pediatric appendectomies on the misdiagnosis of appendicitis in children. Pediatrics.2004;113 :18– 23
- ↵Johnson WG, Rimsza ME. The effects of access to pediatric care and insurance coverage on emergency department utilization. Pediatrics.2004;113 :483– 487
- ↵Kempe A, Beaty BL, Crane LA, et al. Changes in access, utilization, and quality of care after enrollment into a state child health insurance plan. Pediatrics.2005;115 :364– 371
- ↵Szilagyi PG, Dick AW, Klein JD, Shone LP, Zwanziger J, McInerny T. Improved access and quality of care after enrollment in the New York State Children's Health Insurance Program (SCHIP). Pediatrics.2004;113 (5). Available at: www.pediatrics.org/cgi/content/full/113/5/e395
- ↵US Department of Health and Human Services. Delivering on the Promise: Self-Evaluation to Promote Community Living for People with Disabilities Report to the President on Executive Order 13217. Washington, DC: US Department of Health and Human Services; 2002
- ↵Honberg L, McPherson M, Strickland B, Gage JC, Newacheck PW. Assuring adequate health insurance: results of the National Survey of Children With Special Health Care Needs. Pediatrics.2005;115 :1233– 1239
- ↵Berman S, Dolins J, Tang SF, Yudkowsky B. Factors that influence the willingness of private primary care pediatricians to accept more Medicaid patients. Pediatrics.2002;110 :239– 248
- Copyright © 2006 by the American Academy of Pediatrics