Health Care Utilization and Costs of Publicly-Insured Children with Diabetes in California Joyce M. Lee, MD, MPH1,2, Vandana Sundaram, MPH3, Lee Sanders, MD3, Lisa Chamberlain, MD3, and Paul Wise, MD, MPH3 Objective To examine diabetes-related health care utilization and costs for a population-based sample of children with presumed type 1 diabetes (T1D) enrolled in the California Children’s Services program.
Study design Our data source was the California Children’s Services claims data for the period July 1, 2009, to June 30, 2012. We studied a sample of 652 children aged 0-21 years who were continuously enrolled for at least 365 days, had an outpatient visit for T1D, and were taking insulin. Results Compared with the younger age groups, individuals in the 19-21 year age group had the highest rates of hospitalization, T1D-specific bed-days, and emergency department visits. The overall median cost for this population was $7654. The overall median costs per year (and proportion of total costs) were $5603 (59%) for hospitalizations, $58 (0.4%) for emergency department visits, $144 (1.3%) for outpatient utilization, $2930 (23%) for insulin, and $1579 (13%) for blood glucose monitoring supplies. For those who used them, the median cost of pumps was an additional $2162. Conclusion Further studies are needed to provide more insight into patterns of care and adverse health outcomes for children with T1D as they transition into young adulthood. The costs of insulin, glucose monitoring supplies, and pump therapy for children with T1D is substantial and may factor into future policy considerations regarding coverage and cost-sharing with families. (J Pediatr 2015;167:449-54).
D
iabetes represents one of the most common chronic diseases of childhood, and its incidence is increasing in children.1 The majority of children with diabetes have type 1 diabetes (T1D),2 which requires intensive management, including frequent daily blood glucose monitoring and multiple daily injections of insulin (anywhere from 4 to 10 times a day). Furthermore, an increasing proportion of children are using technology, including insulin pumps and continuous glucose monitoring systems, to augment management of T1D.3 Thus, T1D in children represents an excellent paradigm for understanding health care utilization and costs of childhood chronic disease, as well as the impact of medical technology on costs of care. The majority of previous studies have focused almost exclusively on adults, who mostly have type 2 diabetes.4-6 To date, only a few studies have evaluated health utilization patterns and costs of diabetes in children. Those studies focused on privately insured children7 and on inpatient rather than outpatient care,8 were based on small sample sizes or clinic-based samples from tertiary care,9 or were from countries other than the US.10,11 Other studies attributed health care costs to specific events rather than measuring claims paid out for services.12 Consequently, we sought to describe diabetes-related health care utilization and costs (both inpatient and outpatient) for a population-based sample of children with T1D enrolled in the California Children’s Services (CCS) program, a safety net program for children with chronic diseases in the state of California. We hypothesized that health care utilization and costs would be higher for adolescents vs younger children, for minority vs white children, for urban vs rural children, and for children receiving pump therapy vs those not on pump therapy.
Methods CCS, the California safety net Title V program established by the Social Security Act of 1935, provides coordination of care and medical coverage for children with diabetes up to 21 years of age who are enrolled in Medi-Cal (the Medicaid program in California) or Healthy Families (the State Children’s Health Insurance Program in California), or who are uninsured or have private insurance if their healthrelated expenditures exceed a designated level, based on income. From the Child Health Evaluation and Research Unit, Division of General Pediatrics, and Division of Pediatric Our data source was CCS paid claims. We used a dataset provided by the Endocrinology, University of Michigan, Ann Arbor, MI; California Department of Health Care Services that included demographic and Center for Policy, Outcomes and Prevention, 1
2
3
Stanford University, Stanford, CA
CCS ED T1D
California Children’s Services Emergency department Type 1 diabetes
Supported by the Lucile Packard Foundation for Children’s Health (to P.W.). The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2015 Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jpeds.2015.04.067
449
THE JOURNAL OF PEDIATRICS
www.jpeds.com
information, geographic information, diagnoses, procedures, and reimbursement information for all paid claims for every eligible child, as well as information about the CCS-eligible diagnosis and eligibility start and end dates for each child enrolled between July 1, 2009, and June 30, 2012. We identified a sample of children aged 0-21 years with a CCS-eligible diagnosis of diabetes enrolled for a continuous period of at least 365 days who had at least 1 outpatient clinic visit for diabetes (based on an International Classification of Diseases, Ninth Revision code of 250.xx) (n = 7993). We elected to focus on children with insulin-treated diabetes, because this would capture all children with T1D; therefore, we further narrowed the population to 7057 children who were taking insulin. We cannot exclude the possibility that the cohort may include some children with type 2 diabetes, however. Given our aim of describing both utilization and costs, we focused on children enrolled in Medicaid fee-for-service plans, because information about costs for children enrolled in managed care is incomplete; this left us with a sample of 652 children. To identify claims/costs of insulin, blood glucose monitoring, and pump supplies, we used National Drug Codes to capture outpatient pharmacy claims and Healthcare Common Procedure Coding System codes for any insulin/ supplies dispensed in clinic visits (see the Appendices 1-4; available at www.jpeds.com). Children were defined as being on a pump if they had at least 1 outpatient claim for a pump during the study period. We used a descriptive-based approach to assess utilization and costs. We estimated median annualized utilization rates of bed-days and hospitalizations (diabetes-related and non– diabetes-related), and emergency department (ED) visits and outpatient visits (diabetes-related). For costs, we assessed median annualized costs for hospitalizations, ED visits, outpatient visits, insulin, testing supplies, and pumps. Costs were annualized, because children could have been enrolled for anywhere from 1 to 3 years. In sensitivity analyses, we also assessed overall expenditures (not limited to diabetesrelated expenditures) for the cohort. Because we have data for the entire sample of children covered by the CCS program, we did not perform significance testing. For our analyses comparing sex, some children had both “M” and “F” coded as sex on different claims, and these children were coded as “unknown” sex.
Results The majority of children were age 10-14 years (37%) or 1518 years (32%), and there was a preponderance of females (Table I). The majority of children were either Hispanic or white and lived in urban areas. A minority of children used an insulin pump (18%) (Table I). For the overall population, the median annual hospitalization rate was 0.7, the median annual bed-day rate was 2.5 days, the median annual ED visit rate was 0.7 visits, and the median annual outpatient visit rate was 2.7. Table II (available at www.jpeds.com) shows the annualized median inpatient and 450
Vol. 167, No. 2
Table I. Demographic characteristics of children with diabetes in the CCS program for the overall population and stratified by insulin pump use Insulin pump use
Characteristics Age, y* 0-4 5-9 10-14 15-18 19-21 Sex Female Male Unknown Race/ethnicity White Black Hispanic Other Unknown Residence county Rural Urban
Overall (n = 652), n (%)
No pump (n = 534), n (%)
Pump (n = 118), n (%)
36 (5.5) 126 (19.3) 239 (36.7) 206 (31.6) 45 (6.9)
26 (4.9) 97 (18.2) 199 (37.3) 172 (32.2) 40 (7.5)
10 (8.5) 29 (24.6) 40 (33.9) 34 (28.8) 5 (4.2)
274 (42) 245 (37.6) 133 (20.4)
214 (40.1) 213 (39.9) 107 (20)
60 (50.8) 32 (27.1) 26 (22)
237 (36.3) 74 (11.4) 201 (30.8) 120 (18.4) 20 (3.1)
171 (32) 65 (12.2) 178 (33.3) 105 (19.7) 15 (2.8)
66 (55.9) 9 (7.6) 23 (19.5) 15 (12.7) 5 (4.2)
65 (10) 587 (90)
45 (8.4) 489 (91.6)
20 (16.9) 98 (83.1)
*Age for the unique cohort represents the age at the start of the first claim between July 1, 2009, and June 30, 2012.
outpatient utilization rates according to demographic characteristics for the entire sample, with n representing the number of children who had a visit. Compared with the younger age groups, children in the 19- to 21-year age group had the highest rates of hospitalizations, bed-days, and ED visits. Although the median rate of hospitalization was similar in males and females, the median rate of bed-days was slightly higher in females. Compared with children of other races, blacks had higher rates of bed-days and ED visits and had lower rates of outpatient visits. Compared with rural children, urban children had higher rates of beddays and lower rates of outpatient visits. For the overall population, the median (proportion of total costs) annual costs were $5603 (59%) for all hospitalizations, $58 (0.4%) for diabetes-specific ED visits, $144 (1.3%) for diabetes-specific outpatient visits, $2930 (23%) for insulin, and $1579 (13%) for blood glucose monitoring supplies. The overall total median cost for the study population was $7654. In a sensitivity analyses of overall expenditures (not limited to diabetes-related expenditures) for the cohort, the total was $11 624 per child. Figures 1-4 (Figure 2 available at www.jpeds.com) show breakdowns of the annualized median cost rates by age group, sex, race, and pump use. For most age groups, the costs of hospitalization were comparable with the costs of medications, supplies, and insulin pumps; however, for the 19- to 21-year age group, the costs of inpatient hospitalization was nearly double. Overall costs were higher for females compared with males, for black and Hispanic children compared with white children, and for children Lee et al
ORIGINAL ARTICLES
August 2015
Figure 1. Median annualized expenditure rate ($) by age.
with insulin pump therapy compared with those not on pump therapy. For children on insulin pump therapy, the median cost of this therapy was $2162. Compared with children not on pump therapy, pump users had higher costs for insulin, testing, and pump supplies, but not for hospitalizations.
Discussion We found that the costs associated with insulin-treated pediatric diabetes are substantial, with an overall annualized median cost of almost $7700 for children enrolled in the CCS program. In the absence of clinical effectiveness data, which were not available in our dataset, we are unable to assess the cost-effectiveness of diabetes care; however, our estimates can be used to assess the economic burden of diabetes for publicly insured children in the US, and provide estimated costs that can be used in future cost-effectiveness analyses of diabetes interventions. Our data also can serve as important benchmarks for a healthcare system that is shifting to a model of accountable care organizations,13 which may change the coordinated management of children with chronic diseases like diabetes and potentially affect the costs of care.
Our overall cost estimate is lower than that reported for children enrolled in employer-based fee-for-service health plans. Shrestha et al,7 using the 2007 MarketScan commercial claims and encounter database, estimated total excess medical expenditures of $9333 for children under age 19 years with presumed T1D, along with a much lower cost of inpatient hospitalizations compared with our results ($1688 vs $5603). Children enrolled in the CCS program have to demonstrate a certain level of income eligibility; therefore, our estimates may be higher compared with those for children enrolled in private plans, given the association of low socioeconomic status with higher rates of diabetic ketoacidosis.14 An additional important difference is the inclusion of individuals 19-21 years of age in our population, given that CCS eligibility ends at age 21 years. We saw a doubling of annual costs of inpatient hospitalizations for individuals in the 19- to 21-year age group compared with children aged 0-14 years, which also may account for our higher inpatient estimates. During the transition to young adulthood, individuals become more independent in their self-care and move from pediatric to adult diabetes care systems; we speculate that lack of continuity in the system could lead to a greater frequency of adverse events like diabetic ketoacidosis. Future studies are needed to provide insight
Health Care Utilization and Costs of Publicly-Insured Children with Diabetes in California
451
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 167, No. 2
Figure 3. Median annualized expenditure rate ($) by race/ethnicity.
into patterns of care and adverse health outcomes in young adults with diabetes. Our study also highlights the high cost of diabetes supplies and medications, which accounted for a substantial proportion of the total median costs (approximately 40%). Our estimated annual cost of medications was lower than that reported for privately insured children ($2930 vs $3629); nonetheless, the costs of all types of insulin were relatively high. The cost of test strips and supplies was also substantial, with an annualized estimate of $1579, higher than the $863 reported by Shrestha et al.7 To our knowledge, the CCS program pays for all supplies for children with diabetes, and the estimate of Shrestha et al may be lower because some private insurers do not pay for testing supplies. Patients with T1D need a minimum of at least 4-6 test strips per day to maintain reasonable glycemic control, and a recent study showed that even more frequent testing, up to 10 times daily, is associated with optimal glycemic outcomes.15 Despite the large number of glucose meter/test strip manufacturers, the average cost of a test strip is approximately $1, even though the estimated production cost is approximately 10-12 cents per strip.16 In response to these costs, public programs like Medicare are now restricting coverage to particular contracted brands of 452
test strips and meters, and also restricting the quantity of strips,17 although we are unaware of current constraints in the CCS program. An additional unique feature of the present study was our ability to provide a yearly estimate of insulin pump therapy in our cohort, which accounted for an additional $2000 annually. The use of pumps in this population was low (18%) compared with the 30%-49% in a national sample of children with T1D from the T1D Exchange Registry.3 With increasing adoption of technology, the overall costs of care may increase, which insurers and policymakers will need to address over time. Finally, the lowest costs of the program were for outpatient care of children with diabetes. Our figure of $144 per year is lower than the $96 reported by a pediatric clinic at an academic medical center, based on data from 2000-2001.17 Previous studies have highlighted the financial obstacles faced by pediatric endocrinologists given these levels of reimbursement and the additional costs associated with multidisciplinary care, which include not only the costs of physician time, but also costs of employing an interdisciplinary team to provide comprehensive care. Interestingly, in our cohort the median annual number of outpatient visits was 2.7, which Lee et al
ORIGINAL ARTICLES
August 2015
Figure 4. Median annualized expenditure rate ($) by insulin pump use.
is lower than the American Diabetes Association–recommended frequency of quarterly visits. More work is needed to gain insight into the factors contributing to the lower frequency of contacts with the healthcare system. Strengths of our study include the population-based sample, the estimation of actual costs based on paid claims, the assessment of outpatient as well as inpatient care, and the ability to estimate the costs of medications, supplies, and insulin pumps. The study by Ying et al9 provided cost estimates for children with diabetes in 2011, but had significant limitations as a single-center study, only 6 months of follow-up, and ascribed rather than measured costs of healthcare medications/supplies and events. Finally, several international studies also have estimated costs, but these estimates were based on non-US populations which have different payment systems that might not be generalizable to the US population.10,11 We acknowledge some limitations of our study. First, without access to health outcome data, which were not part of our dataset, we could not assess the clinical effectiveness or the cost per quality-adjusted life year. Because early investment in diabetes care may offset the development of costly complications in the future, total annual costs is a limited metric. Second, the use of the International Classification of
Diseases, Ninth Revision code 250.xx does not distinguish between T1D and type 2 diabetes. The vast majority of children with diabetes have T1D; however, we acknowledge that some children in our cohort may have had type 2 diabetes. Third, our study population was drawn from the CCS program, a program for children with chronic diseases in California that provides coverage based on income levels, and thus these costs might not generalize to children enrolled in employerbased health plans. Fourth, we cannot rule out the possibility of dual health insurance coverage for the sample, which might have led to cost underestimates. We did have the ability to identify children with dual coverage for those enrolled in Kaiser plans, but less than 5% of those children had dual coverage. Fifth, we did not have information about the duration of diabetes (ie, new-onset vs long-standing diabetes). Finally, we acknowledge that our results are specific to our population of children enrolled in the CCS program, and might not generalize to children with diabetes covered by programs in other states or covered by alternative forms of insurance. Although some previous studies have reported higher expenditures for children with worse diabetes control or with longer disease duration, we could not examine these factors because this clinical information was not available in our
Health Care Utilization and Costs of Publicly-Insured Children with Diabetes in California
453
THE JOURNAL OF PEDIATRICS
www.jpeds.com
administrative database. Therefore, we could not assess the value of care (ie, cost per quality-adjusted life year) for this population. We also did not account for the indirect costs of lost productivity and additional nonmedical costs. Despite these limitations, however, we are able to provide novel estimates of the costs of care, including the costs of medication, supplies, and insulin pump technology, for a relatively understudied pediatric population in the cost literature. Our study provides important information on the direct medical costs associated with children requiring insulin from a state Title V program. These estimates will inform clinicians, payers, and policymakers about the economic burden of pediatric diabetes, and can be used in policy models to assess the overall costs and cost-effectiveness of diabetes interventions in youth. n Submitted for publication Sep 30, 2014; last revision received Mar 11, 2015; accepted Apr 22, 2015. Reprint requests: Joyce M. Lee, MD, MPH, Associate Professor, Pediatric Endocrinology and Health Services Research, Child Health Evaluation and Research Unit, University of Michigan, 300 NIB, Room 6E18, Campus Box 5456, Ann Arbor, MI 48109-5456. E-mail: [email protected]
References 1. Vehik K, Hamman RF, Lezotte D, Norris JM, Klingensmith G, Bloch C, et al. Increasing incidence of type 1 diabetes in 0- to 17-year-old Colorado youth. Diabetes Care 2007;30:503-9. 2. Liese AD, D’Agostino RB, Hamman RF, Kilgo PD, Lawrence JM, Liu LL, et al. The burden of diabetes mellitus among US youth: prevalence estimates from the SEARCH for Diabetes in Youth Study. Pediatrics 2006; 118:1510-8. 3. Beck RW, Tamborlane WV, Bergenstal RM, Miller KM, DuBose SN, Hall CA. The T1D Exchange clinic registry. J Clin Endocrinol Metab 2012;97:4383-9. 4. Meyers JL, Parasuraman S, Bell KF, Graham JP, Candrilli SD. The highcost, type 2 diabetes mellitus patient: an analysis of managed care administrative data. Arch Public Health 2014;72:6.
454
Vol. 167, No. 2 5. Ward A, Alvarez P, Vo L, Martin S. Direct medical costs of complications of diabetes in the United States: estimates for event-year and annual state costs (USD 2012). J Med Econ 2014;17:176-83. 6. O’Brien JA, Patrick AR, Caro J. Estimates of direct medical costs for microvascular and macrovascular complications resulting from type 2 diabetes mellitus in the United States in 2000. Clin Ther 2003; 25:1017-38. 7. Shrestha SS, Zhang P, Albright A, Imperatore G. Medical expenditures associated with diabetes among privately insured US youth in 2007. Diabetes Care 2011;34:1097-101. 8. Lee JM, Okumura MJ, Freed GL, Menon RK, Davis MM. Trends in hospitalizations for diabetes among children and young adults: United States, 1993-2004. Diabetes Care 2007;30:3035-9. 9. Ying AK, Lairson DR, Giardino AP, Bondy ML, Zaheer I, Haymond MW, et al. Predictors of direct costs of diabetes care in pediatric patients with type 1 diabetes. Pediatr Diabetes 2011;12:177-82. 10. Bachle CC, Holl RW, Strassburger K, Molz E, Chernyak N, Beyer P, et al. Costs of paediatric diabetes care in Germany: current situation and comparison with the year 2000. Diabet Med 2012;29:1327-34. 11. Wirehn AB, Andersson A, Ostgren CJ, Carstensen J. Age-specific direct healthcare costs attributable to diabetes in a Swedish population: a register-based analysis. Diabet Med 2008;25:732-7. 12. Abdelrahaman E, Raghavan S, Baker L, Weinrich M, Winters SJ. Racial difference in circulating sex hormone-binding globulin levels in prepubertal boys. Metabolism 2005;54:91-6. 13. Homer CJ, Patel KK. Accountable care organizations in pediatrics: irrelevant or a game changer for children? JAMA Pediatr 2013;167: 507-8. 14. Lewis KR, Clark C, Velarde MC. Socioeconomic factors associated with pediatric diabetic ketoacidosis admissions in Southern West Virginia. Clin Endocrinol (Oxf) 2014;81:218-21. 15. Miller KM, Beck RW, Bergenstal RM, Goland RS, Haller MJ, McGill JB, et al. Evidence of a strong association between frequency of selfmonitoring of blood glucose and hemoglobin A1c levels in T1D exchange clinic registry participants. Diabetes Care 2013;36:2009-14. 16. Tenderich A. The stinging cost of glucose test strips. http://www. diabetesmine.com/2007/10/the-stinging-co.html. Accessed February 17, 2015. 17. North I. Medicare guidelines enforced for diabetic supplies. http:// www.starherald.com/news/local_news/medicare-guidelines-enforcedfor-diabetic-supplies/image_6a09b1ae-a59f-5f69-8863-1ccb0e9fbf38. html. Accessed February 17, 2015.
Lee et al
August 2015
ORIGINAL ARTICLES
Figure 2. Median annualized expenditure rate ($) by sex.
Health Care Utilization and Costs of Publicly-Insured Children with Diabetes in California
454.e1
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 167, No. 2
Table II. Annualized utilization rates by demographic characteristics Total bed-days Characteristics
n*
Overall 296 Age, y 0-4
Study design Our data source was the California Children’s Services claims data for the period July 1, 2009, to June 30, 2012. We studied a sample of 652 children aged 0-21 years who were continuously enrolled for at least 365 days, had an outpatient visit for T1D, and were taking insulin. Results Compared with the younger age groups, individuals in the 19-21 year age group had the highest rates of hospitalization, T1D-specific bed-days, and emergency department visits. The overall median cost for this population was $7654. The overall median costs per year (and proportion of total costs) were $5603 (59%) for hospitalizations, $58 (0.4%) for emergency department visits, $144 (1.3%) for outpatient utilization, $2930 (23%) for insulin, and $1579 (13%) for blood glucose monitoring supplies. For those who used them, the median cost of pumps was an additional $2162. Conclusion Further studies are needed to provide more insight into patterns of care and adverse health outcomes for children with T1D as they transition into young adulthood. The costs of insulin, glucose monitoring supplies, and pump therapy for children with T1D is substantial and may factor into future policy considerations regarding coverage and cost-sharing with families. (J Pediatr 2015;167:449-54).
D
iabetes represents one of the most common chronic diseases of childhood, and its incidence is increasing in children.1 The majority of children with diabetes have type 1 diabetes (T1D),2 which requires intensive management, including frequent daily blood glucose monitoring and multiple daily injections of insulin (anywhere from 4 to 10 times a day). Furthermore, an increasing proportion of children are using technology, including insulin pumps and continuous glucose monitoring systems, to augment management of T1D.3 Thus, T1D in children represents an excellent paradigm for understanding health care utilization and costs of childhood chronic disease, as well as the impact of medical technology on costs of care. The majority of previous studies have focused almost exclusively on adults, who mostly have type 2 diabetes.4-6 To date, only a few studies have evaluated health utilization patterns and costs of diabetes in children. Those studies focused on privately insured children7 and on inpatient rather than outpatient care,8 were based on small sample sizes or clinic-based samples from tertiary care,9 or were from countries other than the US.10,11 Other studies attributed health care costs to specific events rather than measuring claims paid out for services.12 Consequently, we sought to describe diabetes-related health care utilization and costs (both inpatient and outpatient) for a population-based sample of children with T1D enrolled in the California Children’s Services (CCS) program, a safety net program for children with chronic diseases in the state of California. We hypothesized that health care utilization and costs would be higher for adolescents vs younger children, for minority vs white children, for urban vs rural children, and for children receiving pump therapy vs those not on pump therapy.
Methods CCS, the California safety net Title V program established by the Social Security Act of 1935, provides coordination of care and medical coverage for children with diabetes up to 21 years of age who are enrolled in Medi-Cal (the Medicaid program in California) or Healthy Families (the State Children’s Health Insurance Program in California), or who are uninsured or have private insurance if their healthrelated expenditures exceed a designated level, based on income. From the Child Health Evaluation and Research Unit, Division of General Pediatrics, and Division of Pediatric Our data source was CCS paid claims. We used a dataset provided by the Endocrinology, University of Michigan, Ann Arbor, MI; California Department of Health Care Services that included demographic and Center for Policy, Outcomes and Prevention, 1
2
3
Stanford University, Stanford, CA
CCS ED T1D
California Children’s Services Emergency department Type 1 diabetes
Supported by the Lucile Packard Foundation for Children’s Health (to P.W.). The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2015 Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jpeds.2015.04.067
449
THE JOURNAL OF PEDIATRICS
www.jpeds.com
information, geographic information, diagnoses, procedures, and reimbursement information for all paid claims for every eligible child, as well as information about the CCS-eligible diagnosis and eligibility start and end dates for each child enrolled between July 1, 2009, and June 30, 2012. We identified a sample of children aged 0-21 years with a CCS-eligible diagnosis of diabetes enrolled for a continuous period of at least 365 days who had at least 1 outpatient clinic visit for diabetes (based on an International Classification of Diseases, Ninth Revision code of 250.xx) (n = 7993). We elected to focus on children with insulin-treated diabetes, because this would capture all children with T1D; therefore, we further narrowed the population to 7057 children who were taking insulin. We cannot exclude the possibility that the cohort may include some children with type 2 diabetes, however. Given our aim of describing both utilization and costs, we focused on children enrolled in Medicaid fee-for-service plans, because information about costs for children enrolled in managed care is incomplete; this left us with a sample of 652 children. To identify claims/costs of insulin, blood glucose monitoring, and pump supplies, we used National Drug Codes to capture outpatient pharmacy claims and Healthcare Common Procedure Coding System codes for any insulin/ supplies dispensed in clinic visits (see the Appendices 1-4; available at www.jpeds.com). Children were defined as being on a pump if they had at least 1 outpatient claim for a pump during the study period. We used a descriptive-based approach to assess utilization and costs. We estimated median annualized utilization rates of bed-days and hospitalizations (diabetes-related and non– diabetes-related), and emergency department (ED) visits and outpatient visits (diabetes-related). For costs, we assessed median annualized costs for hospitalizations, ED visits, outpatient visits, insulin, testing supplies, and pumps. Costs were annualized, because children could have been enrolled for anywhere from 1 to 3 years. In sensitivity analyses, we also assessed overall expenditures (not limited to diabetesrelated expenditures) for the cohort. Because we have data for the entire sample of children covered by the CCS program, we did not perform significance testing. For our analyses comparing sex, some children had both “M” and “F” coded as sex on different claims, and these children were coded as “unknown” sex.
Results The majority of children were age 10-14 years (37%) or 1518 years (32%), and there was a preponderance of females (Table I). The majority of children were either Hispanic or white and lived in urban areas. A minority of children used an insulin pump (18%) (Table I). For the overall population, the median annual hospitalization rate was 0.7, the median annual bed-day rate was 2.5 days, the median annual ED visit rate was 0.7 visits, and the median annual outpatient visit rate was 2.7. Table II (available at www.jpeds.com) shows the annualized median inpatient and 450
Vol. 167, No. 2
Table I. Demographic characteristics of children with diabetes in the CCS program for the overall population and stratified by insulin pump use Insulin pump use
Characteristics Age, y* 0-4 5-9 10-14 15-18 19-21 Sex Female Male Unknown Race/ethnicity White Black Hispanic Other Unknown Residence county Rural Urban
Overall (n = 652), n (%)
No pump (n = 534), n (%)
Pump (n = 118), n (%)
36 (5.5) 126 (19.3) 239 (36.7) 206 (31.6) 45 (6.9)
26 (4.9) 97 (18.2) 199 (37.3) 172 (32.2) 40 (7.5)
10 (8.5) 29 (24.6) 40 (33.9) 34 (28.8) 5 (4.2)
274 (42) 245 (37.6) 133 (20.4)
214 (40.1) 213 (39.9) 107 (20)
60 (50.8) 32 (27.1) 26 (22)
237 (36.3) 74 (11.4) 201 (30.8) 120 (18.4) 20 (3.1)
171 (32) 65 (12.2) 178 (33.3) 105 (19.7) 15 (2.8)
66 (55.9) 9 (7.6) 23 (19.5) 15 (12.7) 5 (4.2)
65 (10) 587 (90)
45 (8.4) 489 (91.6)
20 (16.9) 98 (83.1)
*Age for the unique cohort represents the age at the start of the first claim between July 1, 2009, and June 30, 2012.
outpatient utilization rates according to demographic characteristics for the entire sample, with n representing the number of children who had a visit. Compared with the younger age groups, children in the 19- to 21-year age group had the highest rates of hospitalizations, bed-days, and ED visits. Although the median rate of hospitalization was similar in males and females, the median rate of bed-days was slightly higher in females. Compared with children of other races, blacks had higher rates of bed-days and ED visits and had lower rates of outpatient visits. Compared with rural children, urban children had higher rates of beddays and lower rates of outpatient visits. For the overall population, the median (proportion of total costs) annual costs were $5603 (59%) for all hospitalizations, $58 (0.4%) for diabetes-specific ED visits, $144 (1.3%) for diabetes-specific outpatient visits, $2930 (23%) for insulin, and $1579 (13%) for blood glucose monitoring supplies. The overall total median cost for the study population was $7654. In a sensitivity analyses of overall expenditures (not limited to diabetes-related expenditures) for the cohort, the total was $11 624 per child. Figures 1-4 (Figure 2 available at www.jpeds.com) show breakdowns of the annualized median cost rates by age group, sex, race, and pump use. For most age groups, the costs of hospitalization were comparable with the costs of medications, supplies, and insulin pumps; however, for the 19- to 21-year age group, the costs of inpatient hospitalization was nearly double. Overall costs were higher for females compared with males, for black and Hispanic children compared with white children, and for children Lee et al
ORIGINAL ARTICLES
August 2015
Figure 1. Median annualized expenditure rate ($) by age.
with insulin pump therapy compared with those not on pump therapy. For children on insulin pump therapy, the median cost of this therapy was $2162. Compared with children not on pump therapy, pump users had higher costs for insulin, testing, and pump supplies, but not for hospitalizations.
Discussion We found that the costs associated with insulin-treated pediatric diabetes are substantial, with an overall annualized median cost of almost $7700 for children enrolled in the CCS program. In the absence of clinical effectiveness data, which were not available in our dataset, we are unable to assess the cost-effectiveness of diabetes care; however, our estimates can be used to assess the economic burden of diabetes for publicly insured children in the US, and provide estimated costs that can be used in future cost-effectiveness analyses of diabetes interventions. Our data also can serve as important benchmarks for a healthcare system that is shifting to a model of accountable care organizations,13 which may change the coordinated management of children with chronic diseases like diabetes and potentially affect the costs of care.
Our overall cost estimate is lower than that reported for children enrolled in employer-based fee-for-service health plans. Shrestha et al,7 using the 2007 MarketScan commercial claims and encounter database, estimated total excess medical expenditures of $9333 for children under age 19 years with presumed T1D, along with a much lower cost of inpatient hospitalizations compared with our results ($1688 vs $5603). Children enrolled in the CCS program have to demonstrate a certain level of income eligibility; therefore, our estimates may be higher compared with those for children enrolled in private plans, given the association of low socioeconomic status with higher rates of diabetic ketoacidosis.14 An additional important difference is the inclusion of individuals 19-21 years of age in our population, given that CCS eligibility ends at age 21 years. We saw a doubling of annual costs of inpatient hospitalizations for individuals in the 19- to 21-year age group compared with children aged 0-14 years, which also may account for our higher inpatient estimates. During the transition to young adulthood, individuals become more independent in their self-care and move from pediatric to adult diabetes care systems; we speculate that lack of continuity in the system could lead to a greater frequency of adverse events like diabetic ketoacidosis. Future studies are needed to provide insight
Health Care Utilization and Costs of Publicly-Insured Children with Diabetes in California
451
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 167, No. 2
Figure 3. Median annualized expenditure rate ($) by race/ethnicity.
into patterns of care and adverse health outcomes in young adults with diabetes. Our study also highlights the high cost of diabetes supplies and medications, which accounted for a substantial proportion of the total median costs (approximately 40%). Our estimated annual cost of medications was lower than that reported for privately insured children ($2930 vs $3629); nonetheless, the costs of all types of insulin were relatively high. The cost of test strips and supplies was also substantial, with an annualized estimate of $1579, higher than the $863 reported by Shrestha et al.7 To our knowledge, the CCS program pays for all supplies for children with diabetes, and the estimate of Shrestha et al may be lower because some private insurers do not pay for testing supplies. Patients with T1D need a minimum of at least 4-6 test strips per day to maintain reasonable glycemic control, and a recent study showed that even more frequent testing, up to 10 times daily, is associated with optimal glycemic outcomes.15 Despite the large number of glucose meter/test strip manufacturers, the average cost of a test strip is approximately $1, even though the estimated production cost is approximately 10-12 cents per strip.16 In response to these costs, public programs like Medicare are now restricting coverage to particular contracted brands of 452
test strips and meters, and also restricting the quantity of strips,17 although we are unaware of current constraints in the CCS program. An additional unique feature of the present study was our ability to provide a yearly estimate of insulin pump therapy in our cohort, which accounted for an additional $2000 annually. The use of pumps in this population was low (18%) compared with the 30%-49% in a national sample of children with T1D from the T1D Exchange Registry.3 With increasing adoption of technology, the overall costs of care may increase, which insurers and policymakers will need to address over time. Finally, the lowest costs of the program were for outpatient care of children with diabetes. Our figure of $144 per year is lower than the $96 reported by a pediatric clinic at an academic medical center, based on data from 2000-2001.17 Previous studies have highlighted the financial obstacles faced by pediatric endocrinologists given these levels of reimbursement and the additional costs associated with multidisciplinary care, which include not only the costs of physician time, but also costs of employing an interdisciplinary team to provide comprehensive care. Interestingly, in our cohort the median annual number of outpatient visits was 2.7, which Lee et al
ORIGINAL ARTICLES
August 2015
Figure 4. Median annualized expenditure rate ($) by insulin pump use.
is lower than the American Diabetes Association–recommended frequency of quarterly visits. More work is needed to gain insight into the factors contributing to the lower frequency of contacts with the healthcare system. Strengths of our study include the population-based sample, the estimation of actual costs based on paid claims, the assessment of outpatient as well as inpatient care, and the ability to estimate the costs of medications, supplies, and insulin pumps. The study by Ying et al9 provided cost estimates for children with diabetes in 2011, but had significant limitations as a single-center study, only 6 months of follow-up, and ascribed rather than measured costs of healthcare medications/supplies and events. Finally, several international studies also have estimated costs, but these estimates were based on non-US populations which have different payment systems that might not be generalizable to the US population.10,11 We acknowledge some limitations of our study. First, without access to health outcome data, which were not part of our dataset, we could not assess the clinical effectiveness or the cost per quality-adjusted life year. Because early investment in diabetes care may offset the development of costly complications in the future, total annual costs is a limited metric. Second, the use of the International Classification of
Diseases, Ninth Revision code 250.xx does not distinguish between T1D and type 2 diabetes. The vast majority of children with diabetes have T1D; however, we acknowledge that some children in our cohort may have had type 2 diabetes. Third, our study population was drawn from the CCS program, a program for children with chronic diseases in California that provides coverage based on income levels, and thus these costs might not generalize to children enrolled in employerbased health plans. Fourth, we cannot rule out the possibility of dual health insurance coverage for the sample, which might have led to cost underestimates. We did have the ability to identify children with dual coverage for those enrolled in Kaiser plans, but less than 5% of those children had dual coverage. Fifth, we did not have information about the duration of diabetes (ie, new-onset vs long-standing diabetes). Finally, we acknowledge that our results are specific to our population of children enrolled in the CCS program, and might not generalize to children with diabetes covered by programs in other states or covered by alternative forms of insurance. Although some previous studies have reported higher expenditures for children with worse diabetes control or with longer disease duration, we could not examine these factors because this clinical information was not available in our
Health Care Utilization and Costs of Publicly-Insured Children with Diabetes in California
453
THE JOURNAL OF PEDIATRICS
www.jpeds.com
administrative database. Therefore, we could not assess the value of care (ie, cost per quality-adjusted life year) for this population. We also did not account for the indirect costs of lost productivity and additional nonmedical costs. Despite these limitations, however, we are able to provide novel estimates of the costs of care, including the costs of medication, supplies, and insulin pump technology, for a relatively understudied pediatric population in the cost literature. Our study provides important information on the direct medical costs associated with children requiring insulin from a state Title V program. These estimates will inform clinicians, payers, and policymakers about the economic burden of pediatric diabetes, and can be used in policy models to assess the overall costs and cost-effectiveness of diabetes interventions in youth. n Submitted for publication Sep 30, 2014; last revision received Mar 11, 2015; accepted Apr 22, 2015. Reprint requests: Joyce M. Lee, MD, MPH, Associate Professor, Pediatric Endocrinology and Health Services Research, Child Health Evaluation and Research Unit, University of Michigan, 300 NIB, Room 6E18, Campus Box 5456, Ann Arbor, MI 48109-5456. E-mail: [email protected]
References 1. Vehik K, Hamman RF, Lezotte D, Norris JM, Klingensmith G, Bloch C, et al. Increasing incidence of type 1 diabetes in 0- to 17-year-old Colorado youth. Diabetes Care 2007;30:503-9. 2. Liese AD, D’Agostino RB, Hamman RF, Kilgo PD, Lawrence JM, Liu LL, et al. The burden of diabetes mellitus among US youth: prevalence estimates from the SEARCH for Diabetes in Youth Study. Pediatrics 2006; 118:1510-8. 3. Beck RW, Tamborlane WV, Bergenstal RM, Miller KM, DuBose SN, Hall CA. The T1D Exchange clinic registry. J Clin Endocrinol Metab 2012;97:4383-9. 4. Meyers JL, Parasuraman S, Bell KF, Graham JP, Candrilli SD. The highcost, type 2 diabetes mellitus patient: an analysis of managed care administrative data. Arch Public Health 2014;72:6.
454
Vol. 167, No. 2 5. Ward A, Alvarez P, Vo L, Martin S. Direct medical costs of complications of diabetes in the United States: estimates for event-year and annual state costs (USD 2012). J Med Econ 2014;17:176-83. 6. O’Brien JA, Patrick AR, Caro J. Estimates of direct medical costs for microvascular and macrovascular complications resulting from type 2 diabetes mellitus in the United States in 2000. Clin Ther 2003; 25:1017-38. 7. Shrestha SS, Zhang P, Albright A, Imperatore G. Medical expenditures associated with diabetes among privately insured US youth in 2007. Diabetes Care 2011;34:1097-101. 8. Lee JM, Okumura MJ, Freed GL, Menon RK, Davis MM. Trends in hospitalizations for diabetes among children and young adults: United States, 1993-2004. Diabetes Care 2007;30:3035-9. 9. Ying AK, Lairson DR, Giardino AP, Bondy ML, Zaheer I, Haymond MW, et al. Predictors of direct costs of diabetes care in pediatric patients with type 1 diabetes. Pediatr Diabetes 2011;12:177-82. 10. Bachle CC, Holl RW, Strassburger K, Molz E, Chernyak N, Beyer P, et al. Costs of paediatric diabetes care in Germany: current situation and comparison with the year 2000. Diabet Med 2012;29:1327-34. 11. Wirehn AB, Andersson A, Ostgren CJ, Carstensen J. Age-specific direct healthcare costs attributable to diabetes in a Swedish population: a register-based analysis. Diabet Med 2008;25:732-7. 12. Abdelrahaman E, Raghavan S, Baker L, Weinrich M, Winters SJ. Racial difference in circulating sex hormone-binding globulin levels in prepubertal boys. Metabolism 2005;54:91-6. 13. Homer CJ, Patel KK. Accountable care organizations in pediatrics: irrelevant or a game changer for children? JAMA Pediatr 2013;167: 507-8. 14. Lewis KR, Clark C, Velarde MC. Socioeconomic factors associated with pediatric diabetic ketoacidosis admissions in Southern West Virginia. Clin Endocrinol (Oxf) 2014;81:218-21. 15. Miller KM, Beck RW, Bergenstal RM, Goland RS, Haller MJ, McGill JB, et al. Evidence of a strong association between frequency of selfmonitoring of blood glucose and hemoglobin A1c levels in T1D exchange clinic registry participants. Diabetes Care 2013;36:2009-14. 16. Tenderich A. The stinging cost of glucose test strips. http://www. diabetesmine.com/2007/10/the-stinging-co.html. Accessed February 17, 2015. 17. North I. Medicare guidelines enforced for diabetic supplies. http:// www.starherald.com/news/local_news/medicare-guidelines-enforcedfor-diabetic-supplies/image_6a09b1ae-a59f-5f69-8863-1ccb0e9fbf38. html. Accessed February 17, 2015.
Lee et al
August 2015
ORIGINAL ARTICLES
Figure 2. Median annualized expenditure rate ($) by sex.
Health Care Utilization and Costs of Publicly-Insured Children with Diabetes in California
454.e1
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Vol. 167, No. 2
Table II. Annualized utilization rates by demographic characteristics Total bed-days Characteristics
n*
Overall 296 Age, y 0-4
