Objective. To examine the clinical characteristics and health service use of children with high Medicaid expenditures.
Methodology. We examined 1992 Medicaid claims and eligibility files from four states (California, Georgia, Michigan, Tennessee) for children with at least $10 000 billed to Medicaid who obtained Medicaid through the Supplemental Security Income (SSI) Program and a comparison group (matched by age group and gender) of children receiving Medicaid for other reasons. We compared mean expenditures, examined expenses by category, and examined diagnoses associated with at least $10 000 in expenses.
Results. In 1992, Medicaid paid on average ∼$1000 for children with non-SSI Medicaid enrollment. Expenditures for children with SSI were 2.9 to 9.4 times higher, but once the ∼10% of children with high expenditures were excluded, SSI average expenditures were only 1.5 to 2.7 times higher than the non-SSI average. Children with high expenditures are likely to use hospitals and long-term care, and these services account for more than half of the average expenditures. Children with high expenditures and SSI are more likely to have chronic medical conditions than are their peers enrolled in Medicaid but not through SSI.
Conclusions. A small proportion of children, even on SSI, account for very large proportions of Medicaid expenditures. Most children with SSI, despite having relatively severe mental health, physical, or developmental disabilities, have relatively modest Medicaid expenditures.
A relatively few individuals skew the distribution of health care expenditures by consuming a disproportionate share of health care dollars.1,,2 This pattern affects children in general3 and Medicaid populations as well.4–6 Although the disproportionate contribution of high-expenditure individuals to overall expenditures is well documented, few studies analyze the sources of the high expenditures. Children, including children with disabilities insured by Medicaid, are increasingly covered by managed care programs.7,,8 This growth of managed care in part reflects concerns regarding increasing health care costs. Because children with high medical expenditures use such a disproportionate share of health care resources for children, it is important to understand the types of conditions and the types of expenditures that account for these high expenditures.
The Supplemental Security Income (SSI) Program provides cash benefits and (in most states) automatic Medicaid enrollment to children and adolescents with low-income and disabilities. As such, it covers a population more likely to have high medical care expenditures. Congress originally designed the program to provide cash benefits to adults disabled such that they could no longer work. Congress enacted the SSI childhood disability program in 1972 and implemented the program in 1974.9 Since 1989, several policy changes have contributed to a rapid expansion in the program from 275 000 children to >1 000 000 in 1996.10 The rapid growth in the child and adolescent SSI program has caused controversy regarding the purpose of the program, the types of children who have gained access to benefits, and state-to-state variations in implementation of this Federal program.11 Some have questioned whether all children with serious disabilities should receive the same cash benefit and early welfare reform proposals would have limited cash benefits to children with only certain types of disabilities. The final Personal Responsibility and Work Opportunity Act of 1996 (the Welfare Reform Act) maintained cash benefits for all children receiving SSI.
Concerns have also been raised regarding whether children with less severe conditions, especially milder forms of mental retardation or behavioral disorders such as attention deficit hyperactivity disorder (ADHD), have inappropriately obtained eligibility. Although recent data indicate marked growth in numbers of children with ADHD in the program,9,,12 there is little evidence of similar growth in the diagnosis of mental retardation. Despite the changes in ADHD rates, most children and adolescents receiving benefits have physical disabilities and/or severe mental retardation. In 1993, for example, 42% of recipients under age 18 had mental retardation, 36% had chronic physical disorders, and only 22% had mental health conditions other than retardation (including ADHD).13 The 1996 Welfare Reform Act included provisions to restrict the definition of disability and to reevaluate eligibility for some recently enrolled children. These changes will likely slow growth in the SSI population and especially those with less severe disabilities. Finally, although states do vary in the percentages of their child and adolescent populations enrolled in SSI, state poverty levels account for most of this variation in enrollment.14
Studies before the major program expansions indicate that use of expensive resources, especially hospitalization and long-term care, account for more than half of total Medicaid expenditures for adults and children enrolled via SSI.15 Children with SSI enrollment and with Medicaid expenditures >$25 000 in selected states between 1984 and 1986 typically were very young children with high expenditures related to hospital stays or older teens with mental disabilities who were institutionalized.4 Children with SSI enrollment and high and persistent Medicaid expenditures in California in 1983 were generally in institutional settings, whereas those with short-term high costs tended to use acute hospital services.16 Among children with Medicaid enrollment in Washington state, with select tracer conditions, inpatient care is a major contributor to expenditures for children with selected chronic conditions.5
This article describes the clinical characteristics and health service use of children with SSI and high Medicaid expenditures compared with an age and gender matched non-SSI Medicaid-enrolled comparison group. The analytic focus is on the SSI group because of the large increases in program enrollment in recent years and because the program is designed to provide benefits to children who are at high risk of having high medical expenditures. The non-SSI comparison group provides context to the SSI analysis.
We conducted a cross-sectional study of SSI recipients and a matched sample of non-SSI Medicaid enrollees to provide a comparative description of clinical and utilization characteristics of high-cost patients.
The sample includes Medicaid claims and enrollment information for children with SSI enrollment in Georgia, Tennessee, Michigan, and California in 1992 and an age-gender matched comparison group of children with Medicaid coverage for the full calendar year not enrolled in SSI during that year (the non-SSI Medicaid group). We chose these states because they had Medicaid data available for research purposes. These states cover several regions of the country, and include a state with relatively generous Medicaid benefits (California) and a state with limitations in coverage of psychiatric treatments (Georgia). The data come from the contractor to the Health Care Financing Administration who created the master data files as part of the Medicaid tape-to-tape project (Medstat, Santa Barbara, CA).
To be reimbursed for services to Medicaid patients (under fee-for-service plans that were the norm in these states in 1992), health care providers must file claims with the state. All providers must note the following information for each service provided: type of claim (eg, physician, pharmacy, hospital, long-term care facility), recipient number, provider number, diagnoses, procedures, and date of service. Because the primary purpose of the claims data is to ensure provider reimbursement, expenditure data from Medicaid claims are considered relatively reliable. Diagnosis and procedure codes from physician, inpatient, and long-term care claims may be subject to bias because providers will tend to code conditions and procedures that are likely to be reimbursed. Because the population covered by Medicaid is poor and because coverage offered by Medicaid is relatively comprehensive, the Medicaid claims should represent most health care used for children with Medicaid coverage (except for children with other insurance coverage and with capitated care).
Information on the child's Medicaid eligibility and sociodemographic characteristics are contained in eligibility files. We excluded children with capitated insurance coverage (for children with Medicaid through SSI, the highest capitated coverage penetration was 7.5% in California), children with less than a full calendar year of enrollment, children with private insurance, and children who were enrolled in both SSI and non-SSI Medicaid during the calendar year. Children enrolled in capitated care have poor quality Medicaid claims because providers are paid a fixed rate rather than payment by claims. These providers do not need to submit claims to be reimbursed, so that even where Medicaid administration requests claims-like information the quality may be quite poor. Similarly, Medicaid claims likely do not reflect all annual health care provided to children with private insurance or to children who are enrolled for less than a full calendar year. Children can be enrolled in both SSI and non-SSI Medicaid if they transferred from one program to another. We excluded children who transferred eligibility from the analyses to maintain a clean comparison between the SSI and non-SSI Medicaid groups.
We defined a child as having high expenditures if total 1992 Medicaid expenditures were >$10 000 (1992 dollars).
We used Medicaid eligibility codes to determine SSI status and considered children with enrollment in the categorically-needy, blind or disabled, eligibility categories to have SSI. We included children with enrollment in other Medicaid categories (mostly Aid to Families With Dependent Children [AFDC]) in the comparison group.
Based on the information in the Medicaid eligibility files, we coded the date of birth into 3-year age groups and reclassified the four racial/ethnic categories (white, black, other, and not available) into black versus other.
The eligibility file also contains the zip code of the child's residence. We merged the zip code information with 1990 United States Census data to obtain the percent of the zip code area that was urban. We considered any child living in an area with a zip code that was at least 50% urban as living in an urban area and coded indicator variables for urban residence, nonurban residence, and missing information. Systematic differences might account for children having missing data for zip code. For example, children living in institutions might be more likely to have missing zip code information. Michigan data did not contain zip code information for any individuals so the urban variables are not included in the Michigan analyses.
The claims files contain information on the number of long-term care days, institutional care for populations with mental retardation (ICFMR), hospital days, psychiatric hospital days, and number of home health visits. We created seven indicator variables potentially related to greater than average expenditures. They indicate whether a child (in 1992) has: 1) any hospital days; 2) >10 hospital days; 3) >10 long-term care days; 4) any structured nursing facility care; 5) any ICFMR days; 6) any psychiatric hospital days; and 7) any home health visits.
We classified children as having a chronic health condition if they had one or more Medicaid claims for a chronic condition during the 1992 calendar year. The list of conditions considered chronic is comprised of an expanded version of previous categorizations of child health conditions.17,,18 Within the chronic condition categories, we also determined whether a child had a claim for mental retardation and/or for a chronic mental health diagnosis. This approach is conservative, because a child might have a chronic health condition for which treatment was sought, but the claim might be labeled with another diagnosis or purpose for visit, or the child may have received no treatment during 1992.
We first computed the average expenditures by state and Medicaid eligibility group (SSI or non-SSI Medicaid) for the whole sample and separately for children with high and nonhigh Medicaid expenditures. We calculated the percent of total expenditures accounted for by the children with high expenditures by dividing total expenditures of children with high expenditures by total expenditures of all children (again, by state and Medicaid eligibility group).
We next determined the percentage of children with high expenditures who met any of the high health care utilization criteria and who had Medicaid claims for chronic severe conditions by state and Medicaid eligibility group. We determined the more common chronic severe conditions by identifying those with a prevalence (based on medical claims) of at least 10% in any state (separately by Medicaid eligibility group). We then calculated mean expenditures by expenditure category for children with high expenditures and SSI eligibility. In addition, we calculated the percent of total expenditures accounted for by each expenditure category.
We used logistic regression to determine the log odds of whether a child is in the high-expenditure group. Independent variables included age, gender, race/ethnicity, urban residence, missing urban residence data, presence of a chronic condition (physical, mental, and developmental are examined separately), and SSI eligibility. We conducted analyses and present data separately by state because of substantial differences in Medicaid policy, SSI policy, and record keeping between states.
Table 1 presents descriptive information on the SSI population and the comparison population by state in 1992. The average expenditures for the SSI population are notably higher than the average costs for the comparison group. Average total expenditures range from $3238 to $6939 for children with SSI coverage and from $741 to $1114 for children in the comparison group. Children with SSI enrollment have between 2.9 and 9.4 times higher total expenditures than do the children in the comparison group. The differences in average Medicaid expenditures for children once the children with high expenditures are excluded, are much lower (between 1.5- and 2.7-fold difference). Moreover, the average expenditures are quite modest even for the SSI population, a group with known disability, once the children with high expenditures are excluded from analyses (range, $1275–$1748).
The SSI population contains a greater percentage of children with high expenditures, between 7.2% and 12.4% of the SSI population, with only 0.8% to 1.7% in the comparison group. In the SSI group, the high-expenditure population accounts for 63.4% to 81.0% of Medicaid expenditures. Children with high expenditures in the comparison group account for only 14.4% to 28.2% of expenditures. This difference reflects the lower percentage of children with high expenditures in the comparison group as well as the lower average expenditure of the children with high expenditures in this group.
Table 2 presents descriptive information for the children with high expenditures by eligibility group and state. A majority of the children with high expenditures in both the SSI and comparison group had at least one hospital day during the past year. A substantial proportion of children in both the SSI and non-SSI Medicaid groups had at least 10 hospital days and at least 10 long-term care days. As Table 3 will show, these expensive providers account for a large proportion of expenditures for these children. Comparisons of hospital and long-term care use between SSI and non-SSI Medicaid groups within states do not show consistently higher utilization for either type of service. Use of structured nursing facilities and care for the population with retardation (ICFMR) is limited almost exclusively to the SSI population. Levels of ICFMR use are low but expenditures are quite high among the users. Children with high expenditures in the comparison group have a higher level of psychiatric hospital days than do the children enrolled in SSI. Children in the comparison group also have higher percentages of children with chronic psychiatric conditions (Georgia is an exception). Only a small proportion of children with high expenditures have home health care claims, with use of home health much greater among the SSI group (between 11% and 41% in SSI high-expenditure group and between 4% and 18% for the non-SSI high-expenditure group). Children with high expenditures who have SSI eligibility are much more likely to have a claim for mental retardation or chronic physical condition than are children with non-SSI Medicaid.
Table 3 shows mean expenditures by category of use for children with SSI Medicaid eligibility and high expenditures. Inpatient care and long-term care account for more than half of the expenditures among children with high expenditures in all states except Michigan. Among those with inpatient care, the range of hospital expenditures in Michigan (the state with the lowest average) is $1764 to $289 408; in California (the state with the highest average) the range is $3724 to $241 984. Top long-term care expenditures range between $81 504 in Georgia to $328 769 in California. Psychiatric hospital expenditures are quite variable with Michigan having the highest level of 15.2% and Georgia not covering these services. Home health expenditures are quite low on average (they account for between 0.3% and 3.9% of total expenditures for this group). Children with high expenditures and non-SSI Medicaid enrollment have a higher percentage of their expenditures accounted for by hospital and psychiatric hospital expenditures compared with children with SSI Medicaid enrollment. Long-term care expenditures account for a small proportion of the total. For this group of children, the high expenditures are created by inpatient hospitalizations and use of psychiatric hospitals (except in Georgia) with very little use of institutional care.
Table 4 lists the chronic severe childhood conditions that had a prevalence >10% in any state by Medicaid eligibility group. These lists provide more detail than the chronic condition characteristics described in Table 2. Children with SSI eligibility have a much larger list of chronic conditions that meet the threshold for inclusion. Children with SSI eligibility have higher identification of mental retardation. Among chronic physical conditions, children in both Medicaid eligibility groups have asthma at the 10% level in at least one state. Diabetes is a common condition among the children in the non-SSI Medicaid group but not among the children in the SSI Medicaid group. Children with SSI enrollment have a large number of chronic conditions that are less common among the children in the non-SSI group, such as mental retardation, cerebral palsy, spina bifida, and schizophrenic disorders.
Table 5 shows the results of logistic regressions predicting the log odds of a child being in the high expenditure category. As expected, children with SSI coverage are more likely to have high expenditures than are their peers with non-SSI Medicaid, controlling for the presence of a claim for a chronic condition. For each of the chronic condition categories in each state, the likelihood of having high expenditures is higher if the child has a claim for a chronic severe childhood condition with the exception of children with claims for psychiatric disorders in Georgia.
In models parallel to those presented in Table 5, we also examined whether high expenditures in the previous year were predictive of high expenditures in 1992. Having high expenditures in 1991 was positively related to having high expenditures in 1992 and reduced the explanatory power of the SSI variable.
Average 1992 Medicaid expenditures for children in the non-SSI group were less than $1000 in all but one state. The ratio of total Medicaid expenditures of children receiving SSI to children in the non-SSI group varied from 2.9 to 9.4. Once children with high expenditures are excluded from the average, children with SSI enrollment have relatively modest expenditures, approximately twice those of the children in the non-SSI group. Thus, the large majority of children with SSI eligibility are like other poor children with relatively low Medicaid expenditures. The average Medicaid expenditures of the low expenditure group are small in comparison to the average yearly SSI cash payment of ∼$5000 per child.
Among children with high expenditures, however, Medicaid expenditures are substantial. The ∼10% of children with SSI enrollment and high expenditures consume close to 70% of the SSI Medicaid resources. The high level of use of long-term care and hospitals among children with high expenditures and the high expenses associated with these services indicate potential areas for targeted cost reduction. For example, expanded supports for home care might reduce expenditures for families who would choose to care for children at home.
The higher rates of physical and developmental chronic conditions among children with SSI enrollment and high expenditures compared with children with non-SSI enrollment are not surprising, insofar as the SSI program is designed to provide benefits to children with severe disabilities. Children with high expenditures are likely to have high rates of more common childhood chronic health conditions (asthma and mental retardation), less common conditions (paralytic syndromes, epilepsy, anomalies of the nervous system and heart), and need hospital and long-term care services. Children in the non-SSI Medicaid comparison group would likely meet income eligibility for SSI because the maximum eligible income is higher for SSI than AFDC. Because SSI cash benefits are greater than AFDC cash benefits, we would expect that children with AFDC enrollment with chronic conditions meeting SSI diagnostic eligibility would tend to switch their enrollment to that program.
The analysis of psychiatric conditions provided some unexpected results. Children with non-SSI Medicaid have higher rates of having a claim for a chronic psychiatric condition and had more use of psychiatric hospitalizations. The Social Security Administration published new and more inclusive listings of mental health conditions eligible. Nonetheless, for SSI eligibility in 1992, many children with mental health conditions may have faced barriers to SSI enrollment or alternatively, they may have higher rates of mental health conditions of lower severity (and therefore not conditions that qualify for SSI).
Georgia has lower levels of identification of psychiatric conditions among both Medicaid groups than other states. Furthermore, the odds ratio predicting high expenditures indicates that Georgia children with psychiatric conditions have significantly lower likelihood of being in the high-expenditure group. Medicaid coverage in Georgia for psychiatric conditions is lower than in the other states (eg, no coverage of psychiatric hospitalizations). Our results likely reflect mainly the generosity of Georgia's Medicaid coverage for mental health services. Interestingly the elimination of psychiatric hospital expenditures does not seem to result in lower total or inpatient hospital expenditures in Georgia.
These data provide estimates of Medicaid expenditures for both children with high expenditures (mainly but not entirely on SSI) and for other children. As such, these data may help states estimate expenditures as they develop their Child Health Initiative Programs, either through Medicaid expansions or through new programs targeted in other ways to households with higher incomes. These results also suggest that because of the more recent cuts in SSI likely reduce the number of children with less severe illness, the cuts will have limited impact on children with high expenditures. They will therefore likely slightly increase average expenditures.
The data may also help with targeting opportunities to diminish some Medicaid expenditures or to allocate them to more effective or efficient use. For example, the high rates of asthma and other chronic conditions suggest that programs targeted to prevention of asthma hospitalizations in Medicaid populations (especially children with particularly severe cases) could diminish the use of hospital services for their care. Many children with mental retardation use ICFMR facilities. Whether less expensive alternatives to these services can be found is less clear, although expanded community and family-based services may help some children stay out of institutions. The use of case management or other managed care methods for children with identified high expenditures could improve care by helping to prevent hospitalization.
This work was supported in part by Grant MCJ-250634 from the Research Grants Program of the Bureau of Maternal and Child Health, US Department of Health and Human Services (USDHHS) and Cooperative Agreement 18-C-90455/1–01 from the Health Care Financing Administration, USDHHS.
- Received January 19, 1998.
- Accepted April 2, 1998.
Reprint requests to (K.K.) Massachusetts General Hospital, WACC, 715 Fruit St, Boston, MA 02114.
- SSI =
- Supplemental Security Income •
- ADHD =
- attention deficit hyperactivity disorder •
- AFDC =
- Aid to Families with Dependent Children •
- ICFMR =
- institutional care for populations with mental retardation
- ↵Butler JA, Budetti P, McManus MA, Stenmark S, Newacheck PW. Health care expenditures for children with chronic illnesses. In: Hobbs N, Perrin JM, eds. Issues in the Care of Children With Chronic Illness. San Francisco, CA: Jossey-Bass Publishers; 1985
- ↵Ellwood MR, Ku L. Summary and Policy Implications: Analyses of Medicaid Financing for Disabled and High Cost Children. Washington, DC: Office of Family, Community and Long-Term Care Policy, Office of the Assistant Secretary for Planning and Evaluation, US Department of Health and Human Services (USDHHS); December 1990
- Ireys HT,
- Anderson GF,
- Shaffer TJ,
- Neff JM
- ↵Association of State and Territorial Health Officials. Public health strategies for Medicaid managed care. Association for State and Territorial Health Officials Access Rep. 1995;1–8
- ↵National Commission on Childhood Disability. Supplemental Security Income for Children With Disabilities. Washington, DC: US Government Printing Office; 1995
- ↵US General Accounting Office. Social Security: Rapid Rise in Children on SSI Disability Rolls Follows New Regulations. Report to the Honorable Gerald D. Kleczka, House of Representatives. Washington, DC: GAO/HEHS-225; 1994
- ↵Woodward B, Weiser B. Costs soar for children's disability program. Washington Post. Feb 4, 1994:1
- ↵Perrin JM, Kuhlthau K, McLaughlin TJ, Ettner SL, Gortmaker SL. Changing patterns of conditions among children receiving SSI disability benefits. Paper presented at the annual meeting of the Ambulatory Pediatric Association; May 8, 1996; Washington, DC
- ↵Social Security Administration. Children Receiving SSI. Baltimore, MD: December 1995
- Perrin JM,
- Ettner SL,
- McLaughlin TJ,
- Gortmaker SL,
- Bloom SR,
- Kuhlthau K
- ↵Ellwood MR, Herz E. Longitudinal Analysis of High Cost Medicaid Children in California. Lexington, MA: SysteMetrics/McGraw-Hill, Inc; October 1990
- Copyright © 1998 American Academy of Pediatrics