Profile of medical charges for children by health status group and severity level in a Washington State Health Plan.

There is considerable evidence that costs of care vary widely across medical conditions and general health status, and children with certain conditions generate higher medical costs than the average pediatric population (Newacheck and Halfon 1998). There is limited information, however, concerning how these costs vary by health status group or by severity levels within these groups. Health plans also are limited in their ability to identify children with special health care needs for care management strategies. As we enter another time of double-digit health care inflation, there will certainly be efforts to reduce and control costs. An ability to understand and track the costs of specific components of care across pediatric populations will help provide a basis for developing rational care management strategies for cost controls and improved outcomes. There is relatively little information published on medical care costs for children. Ireys et al. (1997) analyzed health care costs for children in a Washington State Medicaid database for 1992. They analyzed the costs of care for certain children with special health care needs (CSHCN) sentinel conditions according to various cost categories, such as hospital, pharmaceutical, inpatient, and physician services, and compared costs by category and with the entire group of enrolled children. In general, all of the condition-based groups they identified were more costly than the entire population. Certain groups, such as those with malignancies, were very costly on an individual basis, but collectively, children with respiratory conditions generated higher overall costs. They also showed that the distribution of costs across cost categories varied considerably for different conditions. Using the same data, Andrews et al. (1997), identified certain very costly conditions that might be used for carve outs and tracking. McCormick et al. (2001), using data from the Medical Expenditure Panel Survey (MEPS) analyzed expenditures for all health care services from any sources during 1996 for children, 0–17 years old. In this study, average annual expenditures for a child with any medical expenses (85.4 percent of the children) were $1,019, and 10 percent of the most expensive children accounted for 69 percent of the total dollars spent on all forms of children's health care. Maynard et al. (2000), with 1994 data from a sample of 67,432 Washington State commercially insured children 2–17 years old, analyzed medical health care plan utilization and health care charges data. They determined that the average per-member, per-year charge was $499. There is some literature on the differential costs between children who have chronic conditions and those who do not. Silber, Gleeson, and Huaqin (1999) described the influence of chronic diseases on resource utilization for common pediatric conditions, and showed that length of inpatient stay and total inpatient charges are higher for common pediatric conditions when the child also has a chronic condition. Several studies have focused on the costs of caring for children with specific chronic health conditions. Guevara et al. 2001, and Chan, Zhan, and Homer 2002, analyzed health service utilization and costs for children with attention deficit hyperactivity disorder, while Lozano et al. 1997 focused on utilization and costs of care for children with asthma. Ringel and Sturm (2001) estimated mental health expenditures for children in 1998. Neff et al. (2001) analyzed the overall costs for the Washington State SSI population in 1992 by broadly defined severity groups, and Kuhlthau et al. (1998), analyzed the costs for SSI enrollees in four states. Neither Neff nor Kuhlthau analyzed the specific components of those costs. A four-state analysis of Medicaid enrollees by the Center for Health Care Strategies (CHCS) demonstrated that children without chronic conditions had average total Medicaid expenses of $37 a month ($444 a year) in comparison to children with chronic or disabling conditions who had average expenditures of $240 per month ($2,880 a year) (Center for Health Care Strategies 2000). None of these studies classified all children by a comprehensive classification of health status groupings and severity, and none except the CHCS study, compared the groups they identified to a control group of predominantly healthy children. In addition, the papers by Ireys, Anderson, Neff, Kulthau, and CHCS studied only Medicaid populations. Recently, 3M Health Services Information Systems and the National Association of Children's Hospitals and Related Institutions (NACHRI) have developed software, Clinical Risk Groups (CRGs), which classifies each individual into a health status group and a severity level, using administrative data (Averill et al. 1999). This software has been developed for risk-adjustment purposes and for tracking subpopulations. A team at NACHRI, with extensive pediatrician input, developed the components of the CRGs that relate to children (Muldoon, Neff, and Gay 1997). The availability of this software now allows for analyses of health care utilization and costs by health status groups at various levels of aggregation, severity level, or a combination of the two. Clinical risk groups have been used to identify and classify children in a medium-sized comprehensive health plan in Washington State (Neff et al. 2002). In this study, CRGs seemed to perform well at identifying children who have chronic health conditions that require interaction with the health care system, such as children with malignancies, cystic fibrosis, diabetes, and attention deficit disorders. It performed reasonably well in identifying children with asthma and behavioral problems, but did not perform well in identifying children with conditions that did not require interactions with the health care system, or had conditions that often are not identified in administrative data, such as those with developmental delay or learning disorders. In this single health plan, covering a mixed, small urban/rural population, 9.5 percent of the children were identified as having chronic conditions. The average numbers of unique medical care encounters per child increased by chronic condition complexity (i.e., health status group) and by severity level. It is, therefore, reasonable to expect that CRGs should perform well in identifying patterns of charges for similarly defined health status groups. The purpose of this article is to demonstrate a replicable methodology for stratifying children in a health plan according to health status groups and severity levels and for evaluating patterns of charges and types of medical care services.