Direct Health Care Costs for Children with Pervasive Developmental Disorders: 1996–2002
- First Online:
- Cite this article as:
- Flanders, S.C., Engelhart, L., Pandina, G.J. et al. Adm Policy Ment Health (2007) 34: 213. doi:10.1007/s10488-006-0098-3
- 106 Views
We compared direct costs of treatment of Pervasive Developmental Disorder (PDD), asthma, and diabetes in children aged 3–17 years. A retrospective, claims-based study was conducted using the California Medicaid (Medi-Cal) database (1996–2002). Seven hundred and thirty-one children with PDD were identified and matched for sex with an equal number of randomly selected children with asthma and diabetes. Mean total health care costs for PDD were two- to threefold higher than for asthma and diabetes post-diagnosis ($4,815 vs. $1,469 vs. $2,404, respectively, P < 0.0001). Children with PDD incur significantly greater health care costs when compared with children with other chronic pediatric diseases.
KeywordsAutismCostHealth careCost of illness
Pervasive Developmental Disorders (PDDs) represent a broad range of disorders, which are characterized by varying degrees of impaired social interaction and communication, and restricted, repetitive, and stereotyped patterns of behavior, interest, and activities. The onset of symptoms typically occurs within the first 3 years of a child’s life, and is associated with interruptions and delays in a variety of developmental domains. Clinically, PDDs include Autistic disorder, Asperger’s disorder, Rett syndrome, childhood disintegrative disorder, and PDD-not otherwise specified (PDD-NOS) (APA, 1994).
Autism has recently emerged as a major focus of public health concern in the US as an increasing number of families and individuals seek educational, social, and health care services to deal with its widespread impact (Newschaffer & Curran, 2003). According to recent epidemiological data, the best estimate for the prevalence of all combined PDDs is believed to be 20–60 per 10,000, with the prevalence of autistic disorder estimated at 10 per 10,000 (Fombonne, 2003). Besides autistic disorder, PDD-NOS represents the majority of other PDD diagnoses identified in prevalence estimates (Fombonne, 2003).
Despite its prevalence and the increasing demand for health services, the economic impact and societal costs associated with the treatment and diagnosis of PDD have not been well described (Järbrink, Fombonne, & Knapp, 2003). Birenbaum, Guyot, and Cohen (1990) evaluated health care expenditures for children and young adults with autism and found that the average annual health care expenditure for this population was over twice that for all American children. They also found that children with autism had hospitalization rates twice the average for other children, and four physician visits annually, slightly above the US average for children.
More recently, Järbrink and Knapp (2001) studied costs related to autism in the UK. Using an assumed prevalence of five cases per 10,000, the estimated annual societal costs exceeded 1 billion UK pounds. Using mid-range population prevalence estimates (15 cases per 10,000), Newschaffer and Curran (2003) estimated the annual cost of autistic disorder in the US to be in the billions of dollars, excluding the additional economic costs associated with other PDDs.
With the growing demand for autism-related services, additional data are needed on the costs of diagnosis and treatment for children with PDD in order to facilitate informed decision-making pertaining to access to health care services and reimbursement issues among policymakers. This study estimates the direct costs of diagnosis and treatment for children with PDDs in a state Medicaid population from 1996 to 2002. Specifically, the primary objective of this retrospective, claims-based study is to examine the direct costs of treatment of children aged 3–17 years with a diagnosis of PDD in comparison with two other chronic pediatric diseases that are associated with high direct costs for health care utilization, diabetes, and asthma. Recent reports indicate that children with asthma or diabetes use significantly more health services and incur greater costs than the general population of children using health services (Icks et al., 2004; Lozano, Fishman, VonKorff, & Hecht, 1997; Ungar et al., 2001). The prevalence of asthma and of diabetes was > 10% in children with high Medicaid expenditures (at least $10,000) in 1992 (Kuhlthau, Perrin, Ettner, McLaughlin, & Gortmaker, 1998). In addition, there is a growing concern about the economic impact of these chronic pediatric illnesses since their prevalence is increasing worldwide, they persist through adulthood, and are associated with substantial burden to health care costs (Druss et al., 2001; Icks et al., 2004; Lozano et al., 1997).
This study describes differences in the utilization of psychiatric and nonpsychiatric health care services among children with PDD, diabetes, or asthma, and determines the period prevalence rates of PDD, diabetes, and asthma for each study year based on the entire state Medicaid population.
The current study utilized a 20% random sample of 2.08 million individuals covered by the California Medicaid (Medi-Cal) program to identify children, ages 3–17, who received a diagnosis of PDD, asthma, or diabetes during the period between January 1, 1996 and June 30, 2002. Sample sizes were not calculated for this analysis, rather they were based on the number of eligible subjects identified within the random sample. Eligible subjects with PDD were identified as having at least one of the following primary or secondary ICD-9 diagnostic codes: autistic disorder (299.0), Asperger’s disorder (299.80), PDD-NOS (299.8), childhood disintegrative disorder (299.10), and Rett syndrome (330.8). Eligible subjects with asthma included those with at least one of the following primary or secondary ICD-9 diagnostic codes for asthma: asthma (493), intrinsic asthma (493.1), chronic obstructive asthma (493.2), asthma, unspecified (493.9). Eligible subjects with diabetes consisted of those with at least one primary or secondary ICD-9 diagnostic code for diabetes mellitus (250.x) and/or specific medications used to treat diabetes. After identifying all eligible PDD subjects, an equal number of children with a diagnosis of asthma or diabetes were then randomly selected from the larger pool of children with these disorders. Children with more than one of these three diagnoses (i.e., PDD, asthma, or diabetes) were excluded from the analysis. Thus, the final sample included three equal groups of mutually exclusive diagnostic groups. All subjects in the dataset were de-identified according to HIPAA requirements and the analysis did not require institutional review board review or approval.
All eligible subjects had at least 90 days of continuous Medi-Cal coverage immediately before their first diagnosis of PDD, asthma, or diabetes as well as a minimum of 180 days after diagnosis. Pre- and post-diagnosis periods were defined as up to 180 days before and 365 days after the first diagnosis, respectively. This timeframe allowed for maximum sample size as well as annualizing of the cost estimate. To accurately compare total direct health care costs, children in the three groups were frequency matched to their length of the post-diagnosis follow-up. Direct health care costs were compared among the three diagnostic groups using the dollar amount paid by Medi-Cal to providers. Medical and prescription cost outcomes included in- and outpatient, as well as psychiatric and nonpsychiatric costs. Several special cost outcomes relevant to PDD were assessed separately, including emergency room visits, home visits, outpatient office visits, speech therapy, pathology/lab services, radiology services, and surgical services. In addition, because of the increased prevalence of autism in males, children in the three diagnostic groups were frequency matched on the basis of sex. All costs are presented as US dollars but were not discounted or adjusted for inflation because of the short time period and the complexity of the resource utilization data within the Medi-Cal sample.
The null hypothesis tested was that there was no significant difference in health care costs among patients with PDD, asthma, or diabetes.
Pre- and post-diagnosis health care costs were compared among the three diagnostic groups using Kruskal–Wallis tests (non-parametric statistical test). An analysis of covariance (ANCOVA) of 12-month post-diagnosis mean costs was also conducted controlling for the respective 6-month pre-diagnosis mean costs as well as age group categories (3–6, 7–12, and 13–17 years), sex, and race. Adjusted least-squares means based on actual costs, ranked costs, and log-transformed costs were used in ANCOVA because many cost distributions were skewed and not normally distributed. No sensitivity analysis was performed. A ≤0.05 alpha level (2-tailed) was used to determine statistical significance with no correction for multiple statistical testing. Pairwise comparisons were not conducted. Period prevalence was computed for each of the years 1996 through 2002 for each of the three diagnostic groups, and estimates were based on the entire 20% Medi-Cal sample of children aged 3–17 years.
Among the three diagnostic groups, demographic characteristics were similar, with the exception of mean age and Hispanic race/ethnicity. Children with asthma were the youngest, with a mean age of 6.3 years, compared with 8.6 years for children with PDD, and 12.1 years for children with diabetes (P < 0.0001). A smaller proportion of children with PDD were Hispanic compared with those with asthma or diabetes (5.8% vs. 44.1% vs. 40%, respectively; P < 0.0001). Just over three-quarters (75.4%, n = 551) of the children in each of the diagnostic groups were male (all groups were frequency-matched to the sex distribution in the PDD group).
Direct Health Care Costs: Pre-diagnosis Period
Mean (median) pre-diagnosis health care costs (US $) for children in each diagnostic group, over a period of 6 months
Total costs (medical + prescription)
Emergency room visits
Outpatient office visits*
Direct Health Care Costs: Post-diagnosis Period
Mean (median) post-diagnosis health care costs (US $) for children in each diagnostic group, over a period of 12 months
Total costs (medical + prescription)
Emergency room visits
Outpatient office visits
In this study, we compared children with PDD with those with asthma or diabetes and found that PDD is more costly, per child, relative to other common pediatric chronic illnesses. Both inpatient or outpatient medical costs and prescription costs were higher for children with PDD when compared with costs from a random sample of children with asthma or diabetes. These cost differences were found for both psychiatric and nonpsychiatric services 6 months before the initial diagnosis, as well as 12 months after the diagnosis.
Mean health care claims, which represent service utilization, were also found to be higher for children with PDD than those of children with asthma or diabetes. Children with PDD utilized more in- and outpatient services and used more prescription medication for both psychiatric and nonpsychiatric services than children with asthma and diabetes.
The higher health care costs in the pre-diagnostic period in children with PDD in the present study could possibly reflect the extensive utilization of health care services to rule out other diagnoses before a diagnosis of PDD is confirmed. Another possible explanation is that children with PDD are generally a more vulnerable group even before diagnosis and tend to have higher medical utilization. For instance, the Medical Expenditure Panel Survey (MEPS) showed that children with disabilities, defined by the presence of a limitation in age-appropriate social roles activities, used many more health care services and incurred ∼fourfold higher health care costs ($2,669 vs. $676) than their counterparts without disabilities (Newacheck, Inkelas, & Kim, 2004). It is important to note that the present findings of higher costs before diagnosis in children with PDD were recently replicated in a privately insured pediatric population (Flanders et al., 2006).
In the present study, costs for outpatient services before diagnosis accounted for the largest proportion (49.8%) of total health care costs for children with PDD when compared with costs incurred for inpatient services or prescription medications (22.1% vs. 28.2%, respectively). Similarly, 1 year after a diagnosis was made, the proportion of outpatient costs (46.9%) continued to be greater for children with PDD than those for inpatient services or prescription claims (26.3% vs. 26.8%, respectively). Nevertheless, the significant utilization of inpatient services demonstrates the need for access to more intensive treatment services for this population.
In unadjusted analyses, direct health care costs for children with PDD were significantly higher compared with children with a diagnosis of asthma or diabetes before and after being diagnosed. Some of these differences may have been a function of baseline levels for the respective cost outcomes. While adjusted psychiatric-related costs continued to be higher in PDD children, nonpsychiatric costs were either similar or higher in the other two groups.
Our findings of high utilization of health care service and expenditure for children with PDD are consistent with those of Birenbaum et al. (1990) who found that average health care expenditures for children with autism were more than twice the average for all American children. Data recently presented by the National Immunization Program and the National Center on Birth Defects and Developmental Disabilities within the Centers for Disease Control and Prevention (CDC) show that in the aggregate, total direct medical costs for privately insured individuals with autism range from $4,000 to $6,000 per year, nearly six times more than for those without autism (Grosse & Shimabukuro, 2005, p. 8). This CDC study also showed that beyond 4 years of age, the direct medical costs for a child or adolescent with autism were nine to ten times more than for a child or adolescent without autism.
Others have also found higher than average health care expenditures for children with special needs. In a secondary analysis of the nationally representative 1999 and 2000 MEPS, Newacheck et al. (2004) looked at the financial burden of childhood disability and found that the 7.3% of American children with disabilities used many more services than their counterparts without disabilities, and as a result, had much higher health care expenditures ($2,669 vs. $676). In another analysis of data from the 2000 MEPS, Newacheck and Kim (2005) examined health care expenditures for children with special health care needs (CSHCN). Results indicated that CSHCN had an average health care expenditure of $2,099 compared with children without special health care needs who had an average expenditure of $628. CSHCN also had more than twice as many physician visits and seven times as many nonphysician visits than other children, and had average out-of-pocket costs for all health care twice that of other children ($352 vs. $174). Average annual expenditures for prescribed medications were 10 times higher ($340 vs. $34) for CSHCN. Our cost results are also comparable to the costs of medical care for children with other common childhood psychiatric disorders such as ADHD (Burd, Klug, Coumbe, & Kerbeshian, 2003; Guevara, Lozano, Wickizer, Mell, & Gephart, 2001; Leibson, Katusic, Barbaresi, Ransom, & O’Brien, 2001). However, our cost estimates may underestimate actual health care expenditures associated with PDD, given that in California, Medi-Cal places some restrictions and exclusions for reimbursing certain specialized procedures, such as neuropsychological and educational testing.
Data from our study are noteworthy since the comparison children had other chronic illnesses that are often associated with greater than average health care utilization and expenditures in children. Indeed, in a cohort study of a health maintenance organization in western Washington state, Lozano et al. (1997) found that children with asthma incurred 88% more costs, filled 2.8 times as many prescriptions, made 65% more nonurgent outpatient visits, and had twice as many inpatient days compared with the general population of children using services. Icks et al. (2004) estimated the direct costs of pediatric diabetes care in Germany to be €66.8 million per year (∼$85 million at today’s exchange rate). Our findings underscore the high health care utilization and costs associated with the diagnosis of PDD as medical expenses for these children outstrip those for children who incur greater than average health care expenditures. Because children with PDD are known to be associated very commonly with sub-normal IQ and other co-morbidities, it may be interesting in future studies to compare children with PDDs with matched children having low IQs without autism.
The period prevalence for PDD in this state Medicaid sample increased 114% during the study period, compared with a 42% decrease in asthma. These estimates are similar to those seen during previous research into the prevalence of full syndrome autism in California by Croen, Grether, Hoogstrate, and Selvin (2002) who found a prevalence of 11.0 per 10,000. During their study period (1987–1994), prevalence increased from 5.8 to 14.9 per 10,000, for an absolute change of 9.1 per 10,000 (157% increase). In a study of service utilization in California (California Health and Human Services Agency, 2003), during the 4-year period from December 1998 to December 2002, there was a net increase of 10,017 individuals (a 97% increase) in the autism caseload within the California Department of Developmental Services. Other studies examining populations outside of California have found similar increases in prevalence rates equal to or greater than those in the California autism study (Barbaresi, Katusic, & Colligan, 2005; Yeargin-Allsopp et al., 2003). However, interpretation of these increases in prevalence must be made with caution since they could reflect shifting diagnostic classifications, a greater awareness of autism, changes in eligibility for services such as special education, or true changes in the prevalence of the disease (Croen et al., 2002; Newschaffer & Curran, 2003).
The striking disparity in the proportion of Hispanic children identified with PDD relative to the other two groups is unexplained. These data could be consistent with under-identification of PDD within some ethnic minority groups, or bona fide differences in prevalence rates. The extent of the disparity in rates needs further investigation. Also, the average age at diagnosis of PDD (8.6 years) is notable. Given that some symptoms of PDD are evident before age 3 years (Charman & Baird, 2002), our data suggest that significant delays in identification exist for many children in these economic strata. Such delays would encourage more systematic efforts for screening and earlier identification.
The present study used an ethnically diverse random sample of a large medical population. We examined costs in multiple categories of health care services, allowing for an assessment of the contribution of individual categories of service to the total costs. Additionally, treatment costs were assessed pre- and post-diagnosis. Some limitations of the study should be noted. Health care claims data used in this study might contain coding errors on diagnosis and possibly costs. The data was collected over a 6-year period of time in which other nondisease factors such as formulary changes, decreases in availability of hospital beds, new technology, etc., may have influenced resource utilization among the population members. These potential factors were not considered. The costs between the study years were not discounted and simply aggregated for statistical comparison. No adjustments were made for co-morbidities, other than the exclusion of children with PDD and co-morbid asthma and diabetes. These may have influenced the cost results reported but should not have affected the overall comparisons between disease groups. In addition, this study was designed to compare direct costs of treatment of children with a diagnosis of PDD with children with diabetes or asthma, and the additive effects of co-morbidities on health care service use were not assessed. Future studies are warranted to address this question. Since multiple statistical tests were conducted and not corrected for statistically in our models, false positive findings may also be possible in our results. Lastly, the population used in this study was Medi-Cal patients only and may not be generally representative, because utilization and costs of services may vary by severity of disease, type of insurance coverage, state, and geographic region. We confirmed recently the present findings in an independent sample of privately insured, multi-state pediatric population (Flanders et al., 2006). The data similarly showed that total costs and number of outpatient claims per child were significantly higher 6 months pre-diagnosis and 12 months post-diagnosis for children with PDD (12-month post-diagnosis total costs: $7,090) than for those with asthma ($2,073) or diabetes ($6,194) (Flanders et al., 2006).
Finally, it should be noted that the overall costs of PDDs extend beyond the direct costs reported here. Educational costs are certainly substantial. Indirect costs include family loss of employment or income and increased out-of-pocket expenses (Curran, Sharples, White, & Knapp, 2001; Järbrink et al., 2003). Nevertheless, the findings of this study have important public policy implications. While others are beginning to document the indirect care costs of PDDs, our findings underscore the financial burden of an array of medical costs related to the diagnosis and treatment of PDDs. These data can inform policymakers as they evaluate the need for changes in health care financing as it relates to children with PDD.
Anna Georgieva, MD, PhD and Remon van den Broek, PhD provided editorial support for this manuscript. Preparation of this article was supported by funding from Janssen Medical Affairs, LLC. Recognition to Jeffrey S. Markowitz of Health Data Analytics for his statistical consultation.