BACKGROUND AND OBJECTIVES:

Expenditures for children with noncomplex chronic diseases (NC-CDs) are related to disease chronicity and resource use. The degree to which specific conditions contribute to high health care expenditures among children with NC-CDs is unknown. We sought to describe patient characteristics, expenditures, and use patterns of children with NC-CDs with the lowest (≤80th percentile), moderate (81–95th percentile), high (96–99th percentile), and the highest (≥99th percentile) expenditures.

METHODS:

In this retrospective cross-sectional study, we used the 2014 Truven Medicaid MarketScan Database for claims from 11 states. We included continuously enrolled children (age <18 years) with NC-CDs (n = 1 563 233). We describe per member per year (PMPY) spending and use by each expenditure group for inpatient services, outpatient services, and the pharmacy for physical and mental health conditions. K-means clustering was used to identify expenditure types for the highest expenditure group.

RESULTS:

Medicaid PMPY spending ranged from $1466 (lowest expenditures) to $57 300 (highest expenditures; P < .001); children in the highest expenditure group were diagnosed with a mental health condition twice as often (72.7% vs 34.1%). Cluster analysis was used to identify 3 distinct groups: 83% with high outpatient mental health expenditures (n = 13 033; median PMPY $18 814), 15% with high inpatient expenditures (n = 2386; median PMPY $92 950), and 1% with high pharmacy expenditures (n = 213; median $325 412). Mental health conditions accounted for half of the inpatient diagnoses in the cluster analysis.

CONCLUSIONS:

One percent of children with the highest expenditures accounted for 20% of Medicaid expenditures in children with NC-CDs; mental health conditions account for a large proportion of aggregate Medicaid spending in children with NC-CDs.

What’s Known on This Subject:

Nearly one-third of children with Medicaid have noncomplex chronic diseases (NC-CDs). Expenditures are related to resource use among those with NC-CDs; however, the degree to which specific conditions contribute to high health care expenditures among these children is unknown.

What This Study Adds:

One percent of children accounted for ∼20% of the annual expenditures in children with NC-CDs. Of these children, 83% had high outpatient mental health expenditures, and approximately half of high-cost inpatient stays were for mental health conditions.

Medicaid is the fastest growing insurer, covering >30 million children. Nearly one-third of children with Medicaid have a noncomplex chronic disease (NC-CD).1 NC-CDs include both physical and mental health conditions, which may or may not be lifelong, follow a relapsing and remitting course, and are generally not progressive.2,3 Examples of common NC-CD diagnoses include asthma, obesity, attention-deficit/hyperactivity disorder, and depression. The severity of NC-CDs varies, resulting in a wide range of health care resource use even among children with the same diagnoses.3 NC-CDs are common among children with Medicaid; however, the factors driving expenditures for patients with NC-CDs with high expenditures are largely unknown.

Physical and mental health care expenditures among children are related to the extent of resource use as well as underlying disease chronicity (ie, nonchronic disease, NC-CD, and complex chronic disease [C-CD]). Inpatient admissions account for high expenditures among children, and previous researchers have suggested that mental health may be a key driver of pediatric inpatient resource use and expenditures.4,6 Mental health conditions are common among children with NC-CDs.1 

The degree to which specific conditions contribute to high health care expenditures among children with NC-CDs is unknown. To address these knowledge gaps, using a large Medicaid claims data set of children with NC-CDs, we (1) examine how patient demographic and clinical characteristics vary between patients with a range of expenditures, (2) describe how health care expenditures and use patterns vary for each expenditure group, and (3) identify clusters of patients with NC-CDs within the highest expenditure group on the basis of patterns of health care expenditures.

In this retrospective cross-sectional study, we used the 2014 Truven Medicaid MarketScan Database (Truven Health Analytics, Ann Arbor, MI). In this data set, deidentified demographic, inpatient, outpatient, and retail pharmacy claims data for children covered by Medicaid (fee for service and managed care) from 11 states are collected. The Office of Research Integrity at Children’s Mercy Hospital deemed this study to be exempt from institutional board review.

We identified 1 563 233 children aged 1 to 18 years who were categorized as having NC-CDs with continuous Medicaid enrollment for at least 11 of the 12 months between January 1, 2014, and December 31, 2014 (Supplemental Fig 1). We included only those children who were continuously enrolled to improve the accuracy of NC-CD classification and annual expenditure estimates. To identify children with NC-CDs, we used the Clinical Risk Group (CRG) version 1.12 (3M Health Information Services, Murray, UT) software to group children into mutually exclusive, hierarchical risk groups based on historical International Classification of Diseases, Ninth Revision (ICD-9), Healthcare Common Procedural Coding System, and Current Procedural Terminology codes.2,3 We included patients within the following CRG categories: single minor chronic disease, minor chronic disease in multiple systems, and significant chronic disease (CRG 3–5). We also included a subset of patients with significant chronic disease in 2 organ systems (CRG 6), a group typically described as containing C-CDs. To differentiate NC-CDs in this group, we included only patients without a dominant chronic condition. Dominant chronic conditions are serious medical conditions that are expected to result in a progressive deterioration of health3 in combination with technology dependence and functional status limitations.7 Example conditions of coincident asthma and attention-deficit/hyperactivity disorder (CRG 6 with no dominant chronic condition) would fit our definition of NC-CDs.

Demographic and clinical characteristics included age, sex, race and/or ethnicity, CRG classification, and the presence of mental health conditions. Health care use included inpatient admission, outpatient visits (including visits to the primary care physician, dentist, psychiatrist, pediatric subspecialist, emergency department, and home health visits), retail or mail-order pharmacy services, durable medical equipment, and other services (ie, testing, therapy, charges for drug infusion, and outpatient visits not otherwise specified). Use and costs from an emergency department visit that resulted in hospitalization were attributed to the inpatient admission. Multiple outpatient claims on a single day were recorded as a single use of outpatient resources; however, charges accrued from single visits were captured individually. Discharge diagnoses were classified by using the Centers for Medicare and Medicaid Services Severity Diagnosis-Related Groups (DRGs), in which mutually exclusive ICD-9 codes are compiled into Major Diagnostic Categories (eg, 19: mental diseases and disorders).8 Prescriptions were classified by using Truven’s Red Book.9 

Our primary outcome was expenditures for inpatient and outpatient services; we grouped patients categorically, similarly to previous work,10 in the lowest (≤80th percentile), moderate (81–95th percentile), high (96–99th percentile), and the highest (≥99th percentile) expenditure groups. Standardized expenditures for each service were calculated on the basis of fee-for-service claims by using a previously developed method10,11 and were reported as aggregated and per member per year (PMPY) expenditures. PMPY expenditures were calculated as the total expenditures within an expenditure group divided by the total number of patient-years attributed to the same expenditure group.

Categorical variables were summarized by using frequencies and percentages, and continuous variables were summarized by using medians with interquartile ranges (IQRs) for nonnormally distributed data. Univariate comparisons of demographic and payment information across groups were made by using a χ2 test for association for categorical variables and the Kruskal-Wallis test for continuous data.

For children in the highest expenditure group, we used K-means clustering to assign enrollees to 1 of 3 groups based on similar inpatient service, pharmacy, mental health service, and outpatient specialist payments. K-means clustering is used to partition enrollees into k clusters (in which k represents the number of clusters) such that each enrollee belongs to the cluster with the nearest mean payment pattern. Clustering is a methodology used to identify comparable groupings of observations (eg, enrollees) based on multiple characteristics (eg, payment across the pharmacy and inpatient, mental health, and outpatient services).12 Distinct groups are identifiable when there is significant similarity in the characteristics among observations within a group and easily recognizable differences across groups. The number of clusters (or groups) is determined through an examination of the Scree Plot. Scree Plots allow for a visual demonstration of the number of factors that explain most of the variability. In this case, variability in diagnostic spending was explained by 3 factors (ie, 3 spending groups). The addition of more factors, or groups, explained less and less variability. All analyses were performed by using SAS version 9.4 (SAS Institute, Inc, Cary, NC), and P values <.01 were considered statistically significant.

We included 1 563 233 children with NC-CDs, representing 30.6% of children and 38.8% of expenditures for continuously enrolled children in this Medicaid database. There were 1 250 586 children in the lowest, 234 486 children in the moderate, 62 529 children in the high, and 15 632 children in the highest expenditure category (Table 1). In bivariate analyses in which we compared children in the lowest expenditure group to children in the highest expenditure group, a higher percentage of children in the highest expenditure group were adolescents (48.8% vs 33.0%; P < .001), enrolled in fee-for-service Medicaid (42.8% vs 32.1%; P < .001), classified within CRG 6 (34.5% vs 8.6%; P < .001), diagnosed with a mental health condition (72.7% vs 34.1%; P < .001), and diagnosed with co-occurring physical and mental health conditions (43.1% vs 11.8%; P < .001).

TABLE 1

Patient Sociodemographic and Clinical Characteristics Among Children With NC-CDs Enrolled in Medicaid in 11 States by Expenditure Category

TotalExpenditure Groupa
Lowest (≤80%)Moderate (81%–95%)High (96%–99%)Highest (≥99%)
Children, n 1 563 233 1 250 586 234 486 62 529 15 632 
Patient characteristics, n (%)      
 Age      
  11–23 mo 53 318 (3.4) 33 288 (2.7) 15 285 (6.5) 3900 (6.2) 845 (5.4) 
  2–5 y 306 912 (19.6) 243 080 (19.4) 51 910 (22.1) 10 147 (16.2) 1775 (11.4) 
  6–12 y 678 641 (43.4) 561 707 (44.9) 89 579 (38.2) 21 977 (35.1) 5378 (34.4) 
  13–18 y 524 362 (33.5) 412 512 (33.0) 77 711 (33.1) 26 505 (42.4) 7634 (48.8) 
 Sex      
  Male 873 045 (55.8) 696 936 (55.7) 131 225 (56.0) 35 312 (56.5) 9572 (61.2) 
  Female 690 188 (44.2) 553 651 (44.3) 103 260 (44.0) 27 217 (43.5) 6060 (38.8) 
 Race and/or ethnicity      
  White 758 013 (48.5) 590 103 (47.2) 126 421 (53.9) 33 226 (53.1) 8263 (52.9) 
  African American 516 592 (33.0) 425 460 (34.0) 68 297 (29.1) 18 402 (29.4) 4433 (28.4) 
  Hispanic 124 362 (8.0) 104 108 (8.3) 16 084 (6.9) 3405 (5.4) 765 (4.9) 
  Other 164 266 (10.5) 130 916 (10.5) 23 683 (10.1) 7496 (12.0) 2171 (13.9) 
 Payer      
  Managed care 1 062 812 (68.0) 849 697 (67.9) 161 727 (69.0) 42 443 (67.9) 8945 (57.2) 
  Fee for service 500 421 (32.0) 400 890 (32.1) 72 758 (31.0) 20 086 (32.1) 6687 (42.8) 
 CRG      
  3 (single minor chronic) 451 585 (28.9) 397 023 (31.7) 47 067 (20.1) 6626 (10.6) 869 (5.6) 
  4 (minor chronic in >1 system) 48 978 (3.1) 36 171 (2.9) 10 727 (4.6) 1877 (3.0) 203 (1.3) 
  5A (single moderate chronic) 759 705 (48.6) 633 348 (50.6) 99 130 (42.3) 22 659 (36.2) 4568 (29.2) 
  5B (single dominant chronic) 122 765 (7.9) 76 856 (6.1) 28 453 (12.1) 12 857 (20.6) 4599 (29.4) 
  6A (nondominant significant chronic in >1 system) 180 200 (11.5) 107 189 (8.6) 49 108 (20.9) 18 510 (29.6) 5393 (34.5) 
 Mental health condition      
  Mental health 377 326 (24.1) 278 908 (22.3) 72 903 (31.1) 20 883 (33.4) 4632 (29.6) 
  Physical 940 230 (60.1) 824 679 (65.9) 93 922 (40.1) 17 369 (27.8) 4260 (27.3) 
  Mental health and physical 245 677 (15.7) 147 000 (11.8) 67 660 (28.9) 24 277 (38.8) 6740 (43.1) 
TotalExpenditure Groupa
Lowest (≤80%)Moderate (81%–95%)High (96%–99%)Highest (≥99%)
Children, n 1 563 233 1 250 586 234 486 62 529 15 632 
Patient characteristics, n (%)      
 Age      
  11–23 mo 53 318 (3.4) 33 288 (2.7) 15 285 (6.5) 3900 (6.2) 845 (5.4) 
  2–5 y 306 912 (19.6) 243 080 (19.4) 51 910 (22.1) 10 147 (16.2) 1775 (11.4) 
  6–12 y 678 641 (43.4) 561 707 (44.9) 89 579 (38.2) 21 977 (35.1) 5378 (34.4) 
  13–18 y 524 362 (33.5) 412 512 (33.0) 77 711 (33.1) 26 505 (42.4) 7634 (48.8) 
 Sex      
  Male 873 045 (55.8) 696 936 (55.7) 131 225 (56.0) 35 312 (56.5) 9572 (61.2) 
  Female 690 188 (44.2) 553 651 (44.3) 103 260 (44.0) 27 217 (43.5) 6060 (38.8) 
 Race and/or ethnicity      
  White 758 013 (48.5) 590 103 (47.2) 126 421 (53.9) 33 226 (53.1) 8263 (52.9) 
  African American 516 592 (33.0) 425 460 (34.0) 68 297 (29.1) 18 402 (29.4) 4433 (28.4) 
  Hispanic 124 362 (8.0) 104 108 (8.3) 16 084 (6.9) 3405 (5.4) 765 (4.9) 
  Other 164 266 (10.5) 130 916 (10.5) 23 683 (10.1) 7496 (12.0) 2171 (13.9) 
 Payer      
  Managed care 1 062 812 (68.0) 849 697 (67.9) 161 727 (69.0) 42 443 (67.9) 8945 (57.2) 
  Fee for service 500 421 (32.0) 400 890 (32.1) 72 758 (31.0) 20 086 (32.1) 6687 (42.8) 
 CRG      
  3 (single minor chronic) 451 585 (28.9) 397 023 (31.7) 47 067 (20.1) 6626 (10.6) 869 (5.6) 
  4 (minor chronic in >1 system) 48 978 (3.1) 36 171 (2.9) 10 727 (4.6) 1877 (3.0) 203 (1.3) 
  5A (single moderate chronic) 759 705 (48.6) 633 348 (50.6) 99 130 (42.3) 22 659 (36.2) 4568 (29.2) 
  5B (single dominant chronic) 122 765 (7.9) 76 856 (6.1) 28 453 (12.1) 12 857 (20.6) 4599 (29.4) 
  6A (nondominant significant chronic in >1 system) 180 200 (11.5) 107 189 (8.6) 49 108 (20.9) 18 510 (29.6) 5393 (34.5) 
 Mental health condition      
  Mental health 377 326 (24.1) 278 908 (22.3) 72 903 (31.1) 20 883 (33.4) 4632 (29.6) 
  Physical 940 230 (60.1) 824 679 (65.9) 93 922 (40.1) 17 369 (27.8) 4260 (27.3) 
  Mental health and physical 245 677 (15.7) 147 000 (11.8) 67 660 (28.9) 24 277 (38.8) 6740 (43.1) 
a

P < .001 for all comparisons.

The median PMPY spending for children with NC-CDs was $3208 and varied by expenditure group (all P < .001), ranging from $1466 in the lowest expenditure group to $57 300 in the highest expenditure group (Table 2). The greatest differences between the lowest and highest expenditure groups were for inpatient care, outpatient mental health services, and the pharmacy. For inpatient care, PMPY spending within the highest expenditure group substantially exceeded the PMPY in the lowest expenditure group ($20 183 vs $6). Compared with children in the lowest expenditure group, PMPY spending for children in the highest expenditure group was >70 times greater for outpatient mental health services ($16 018 vs $220) and >80 times greater for the outpatient pharmacy ($10 160 vs $123).

TABLE 2

PMPY Expenditures Among Children With NC-CDs Enrolled in Medicaid in 11 States by Expenditure Category

PMPY ExpendituresTotal, $Expenditure Groupa
Lowest (≤80%), $Moderate (81%–95%), $High (96%–99%), $Highest (≥99%), $
Total 3208 1466 5851 14 612 57 300 
Inpatient 392 422 3043 20 183 
Outpatient 2494 1337 4932 10 381 26 957 
 Mental health 815 220 1641 5796 16 018 
 Primary care physician 301 266 453 432 357 
 Subspecialist physician 298 168 447 1054 5408 
 Emergency department 255 188 518 567 406 
 Pharmacy 322 123 497 1188 10 160 
 Dental 220 200 320 256 221 
 Durable medical equipment 61 50 98 129 172 
 Home health 21 74 1656 
 Other 523 244 1450 2073 2720 
PMPY ExpendituresTotal, $Expenditure Groupa
Lowest (≤80%), $Moderate (81%–95%), $High (96%–99%), $Highest (≥99%), $
Total 3208 1466 5851 14 612 57 300 
Inpatient 392 422 3043 20 183 
Outpatient 2494 1337 4932 10 381 26 957 
 Mental health 815 220 1641 5796 16 018 
 Primary care physician 301 266 453 432 357 
 Subspecialist physician 298 168 447 1054 5408 
 Emergency department 255 188 518 567 406 
 Pharmacy 322 123 497 1188 10 160 
 Dental 220 200 320 256 221 
 Durable medical equipment 61 50 98 129 172 
 Home health 21 74 1656 
 Other 523 244 1450 2073 2720 

Other outpatient expenditures include testing, therapy, treatment, charges for drug infusion, and outpatient visits not otherwise specified.

a

P < .001 for all comparisons.

Three specific patient types within the highest expenditure group were identified by using cluster analysis: patients with a high outpatient mental health expenditure (N = 13 033; 83% of the highest expenditure group; median PMPY $18 814), patients with a high inpatient expenditure (N = 2386; 15% of the highest expenditure group; median PMPY $92 950), and patients with a high pharmacy expenditure (N = 213; 1% of the highest expenditure group; median $325 412; Table 3).

TABLE 3

Predominant Drivers of High Expenditures Among Children With NC-CDs Enrolled in Medicaid in 11 States for the Highest Expenditure Group (≥99%)

High Expenditure ClusterNPrimary Drivers of Costa
Inpatient, $Subspecialist Physician, $Mental Health Service, $Pharmacy, $
Outpatient mental health expenditures 13 033 7059 6260 18 814 6649 
Inpatient expenditures 2386 92 950 1064 1895 1038 
Pharmacy expenditures 213 4603 950 426 325 412 
High Expenditure ClusterNPrimary Drivers of Costa
Inpatient, $Subspecialist Physician, $Mental Health Service, $Pharmacy, $
Outpatient mental health expenditures 13 033 7059 6260 18 814 6649 
Inpatient expenditures 2386 92 950 1064 1895 1038 
Pharmacy expenditures 213 4603 950 426 325 412 
a

Median PMPY expenditures; P < .001 for all comparisons.

Among children who were within the highest expenditure group because of high inpatient expenditures, 5 of the top 10 DRGs were related to mental health, namely, psychosis, behavioral and developmental disorders, depressive neuroses, nondepressive neuroses, and substance abuse (Supplemental Table 5). However, the length of stay for these 5 mental health conditions generally exceeded that for physical conditions, and the costs per day were substantially lower. In contrast, for children within the highest expenditure group secondary to high pharmacy expenditures, only 1 of the 10 most expensive classes of medications were psychotropic medications, namely aripiprazole, with no generic alternative in 2014 (Supplemental Table 6). The cost per claim for this atypical antipsychotic was $144, which is substantially lower than the next most expensive claim for omalizumab ($4030). Furthermore, the most common medication classes in the highest pharmacy expenditures group were growth hormones, recombinant factor medications, and other specialty pharmacy services (eg, immune-modulating medications).

Total Medicaid expenditures equaled $5 billion for children with NC-CDs and varied by expenditure category (Supplemental Table 7). Fifty-two percent of the total inpatient expenditures and 11% of the total outpatient costs were for the 1% of children in the highest expenditure group (ie, ≥99th percentile of expenditures). In addition, health care use for children with NC-CDs varied significantly by expenditure category (all P < .001; Table 4). Only 3.6% of all children had costs directed to inpatient use, although 43.7% in the highest expenditure category were hospitalized. Although most of these children were hospitalized only once, the median length of stay for children was 9 times greater within the highest expenditure category than in the lowest expenditure category (9 days versus 1 day).

TABLE 4

Health Care Use Among Children With NC-CDs Enrolled in Medicaid in 11 States by Expenditure Category

TotalExpenditure Groupa
Lowest (≤80%)Moderate (81%–95%)High (96%–99%)Highest (≥99%)
Inpatient      
 Patients, n (%)b 56 563 (3.6) 3673 (0.3) 24 630 (10.5) 21 432 (34.3) 6828 (43.7) 
 Discharges, median (IQR)c 1 (1–1) 1 (1–1) 1 (1–1) 1 (1–1) 1 (1–2) 
 Hospital d, median (IQR)c 3 (2–5) 1 (1–2) 2 (1–3) 4 (2–6) 9 (5–20) 
Outpatient      
 Mental health      
  Patients, n (%)b 612 259 (39.2) 414 554 (33.1) 141 938 (60.5) 44 721 (71.5) 11 046 (70.7) 
  Claims, median (IQR)c 8 (3–25) 5 (2–12) 28 (9–48) 78 (30–115) 172 (31–292) 
 Primary care physician      
  Patients, n (%)b 1 267 668 (81.1) 997 984 (79.8) 204 717 (87.3) 52 464 (83.9) 12 503 (80.0) 
  Claims, median (IQR)c 4 (2–7) 4 (2–7) 6 (3–10) 5 (2–10) 4 (2–8) 
 Subspecialty physician      
  Patients, n (%)b 1 045 031 (66.9) 791 936 (63.3) 189 013 (80.6) 51 080 (81.7) 13 002 (83.2) 
  Claims, median (IQR)c 3 (2–6) 3 (2–5) 5 (2–9) 6 (3–13) 7 (3–20) 
 Emergency department      
  Patients, n (%)b 620 929 (39.7) 447 642 (35.8) 132 385 (56.5) 33 934 (54.3) 6968 (44.6) 
  Claims, median (IQR)c 1 (1–2) 1 (1–2) 2 (1–3) 2 (1–3) 2 (1–3) 
 Pharmacy      
  Patients, n (%)b 1 380 453 (88.3) 1 084 317 (86.7) 221 987 (94.7) 59 410 (95.0) 14 739 (94.3) 
  Claims, median (IQR)c 9 (4–16) 8 (4–14) 14 (6–25) 17 (8–32) 21 (9–40) 
 Dental      
  Patients, n (%)b 926 459 (59.3) 732 793 (58.6) 147 913 (63.1) 36 833 (58.9) 8920 (57.1) 
  Claims, median (IQR)c 8 (5–11) 8 (5–11) 8 (5–13) 8 (4–12) 7 (4–12) 
 Durable medical equipment      
  Patients, n (%)b 271 235 (17.4) 190 677 (15.2) 58 432 (24.9) 17 389 (27.8) 4737 (30.3) 
  Claims, median (IQR)c 2 (2–4) 2 (2–3) 3 (1–4) 3 (2–5) 3 (2–5) 
 Home health      
  Patients, n (%)b 7241 (0.5) 1736 (0.1) 2261 (1.0) 1710 (2.7) 1534 (9.8) 
  Claims, median (IQR)c 3 (1–13) 1 (1–3) 3 (1–5) 4 (1–21) 22.5 (4–284) 
TotalExpenditure Groupa
Lowest (≤80%)Moderate (81%–95%)High (96%–99%)Highest (≥99%)
Inpatient      
 Patients, n (%)b 56 563 (3.6) 3673 (0.3) 24 630 (10.5) 21 432 (34.3) 6828 (43.7) 
 Discharges, median (IQR)c 1 (1–1) 1 (1–1) 1 (1–1) 1 (1–1) 1 (1–2) 
 Hospital d, median (IQR)c 3 (2–5) 1 (1–2) 2 (1–3) 4 (2–6) 9 (5–20) 
Outpatient      
 Mental health      
  Patients, n (%)b 612 259 (39.2) 414 554 (33.1) 141 938 (60.5) 44 721 (71.5) 11 046 (70.7) 
  Claims, median (IQR)c 8 (3–25) 5 (2–12) 28 (9–48) 78 (30–115) 172 (31–292) 
 Primary care physician      
  Patients, n (%)b 1 267 668 (81.1) 997 984 (79.8) 204 717 (87.3) 52 464 (83.9) 12 503 (80.0) 
  Claims, median (IQR)c 4 (2–7) 4 (2–7) 6 (3–10) 5 (2–10) 4 (2–8) 
 Subspecialty physician      
  Patients, n (%)b 1 045 031 (66.9) 791 936 (63.3) 189 013 (80.6) 51 080 (81.7) 13 002 (83.2) 
  Claims, median (IQR)c 3 (2–6) 3 (2–5) 5 (2–9) 6 (3–13) 7 (3–20) 
 Emergency department      
  Patients, n (%)b 620 929 (39.7) 447 642 (35.8) 132 385 (56.5) 33 934 (54.3) 6968 (44.6) 
  Claims, median (IQR)c 1 (1–2) 1 (1–2) 2 (1–3) 2 (1–3) 2 (1–3) 
 Pharmacy      
  Patients, n (%)b 1 380 453 (88.3) 1 084 317 (86.7) 221 987 (94.7) 59 410 (95.0) 14 739 (94.3) 
  Claims, median (IQR)c 9 (4–16) 8 (4–14) 14 (6–25) 17 (8–32) 21 (9–40) 
 Dental      
  Patients, n (%)b 926 459 (59.3) 732 793 (58.6) 147 913 (63.1) 36 833 (58.9) 8920 (57.1) 
  Claims, median (IQR)c 8 (5–11) 8 (5–11) 8 (5–13) 8 (4–12) 7 (4–12) 
 Durable medical equipment      
  Patients, n (%)b 271 235 (17.4) 190 677 (15.2) 58 432 (24.9) 17 389 (27.8) 4737 (30.3) 
  Claims, median (IQR)c 2 (2–4) 2 (2–3) 3 (1–4) 3 (2–5) 3 (2–5) 
 Home health      
  Patients, n (%)b 7241 (0.5) 1736 (0.1) 2261 (1.0) 1710 (2.7) 1534 (9.8) 
  Claims, median (IQR)c 3 (1–13) 1 (1–3) 3 (1–5) 4 (1–21) 22.5 (4–284) 
a

P < .001 for all comparisons.

b

Percent with charge for service line.

c

Number among children with a service-line charge.

The use of outpatient mental health services was also 2.1 times greater among children within the highest expenditure group (70.7% vs 33.1%) compared with the lowest expenditure group. The frequency of claims for outpatient mental health services far exceeded the claims for any other outpatient care. Children in the highest expenditure category for outpatient mental health services had a 34.4 times greater median number of claims than children in the lowest expenditure group (172 vs 5). Similarly, the median number of pharmacy claims for children in the highest expenditure group was 2.6 times that of children in the lowest expenditure group (21 vs 8).

In this study of 1 563 233 children with NC-CDs and Medicaid insurance, we describe that 1% of children (ie, those in the ≥99th percentile of expenditures) accounted for ∼20% of the annual expenditures for children with NC-CDs. The median PMPY spending in children with NC-CDs was $3208 but varied between $1466 in the lowest expenditure group and $57 300 in the highest expenditure group. The greatest differences between the lowest and highest expenditure groups were for inpatient care, outpatient mental health services, and the pharmacy. Our results reveal that these services are likely the primary drivers of the highest expenditures among children with NC-CDs.

In a cluster analysis of heath care spending in children in the highest 1% of expenditures, we placed children with similar spending patterns together; 83% of children had high expenditures associated with outpatient mental health care, 15% of children had high expenditures associated with high inpatient, and 1% of children had high expenditures associated with high pharmacy expenditures. Although all children included in the cluster analysis have disproportionately high expenditures compared with all children with NC-CDs, there is considerable variation in this group. For instance, children who were in the highest expenditure category because of their frequent use of outpatient mental health services ($18 814) spent much less than children with high expenditures due to inpatient ($92 950) or pharmacy services ($325 412).

Our findings are consistent with previous work in which researchers describe substantial differences in expenditures based on increasing medical complexity among children with Medicaid.10,13 We describe that a plurality of children with the highest expenditures have 3 or more chronic conditions (consistent with C-CD classification), and larger proportions of expenditures within groups of children with the highest expenditure category are due to inpatient services and outpatient mental health service use. With our analysis, we add to previous work by describing that the provision of outpatient mental health services and pharmacy services account for higher proportion of the cost for the population of children with NC-CDs compared with all children covered by Medicaid.

We also describe that half of the diagnoses for hospitalizations for which children with NC-CDs were placed among the most expensive were mental health conditions. It is notable to mention that the children with the highest-cost NC-CDs with mental health hospitalizations had a much longer length of stay but a much lower cost per day compared with those hospitalized for physical diagnoses. These findings are consistent with a previous study that revealed that the overall cost per hospital day was less for hospitalizations with a psychiatric diagnosis compared with those without among children hospitalized in 33 tertiary-care children’s hospitals.4 Together, these findings are used to support future research in which authors identify the predictors of long lengths of hospital stay for children with psychiatric disorders, such as poor care coordination with outpatient mental health care.14 

Although pharmacy services account for higher proportions of the cost for children with NC-CDs with the highest expenditures, the majority of medications that are placed in this category are those prescribed for few patients and that are largely separate from expenditures related to mental health conditions. For aripiprazole, there were 13 262 claims for 1655 patients ($144 per claim), but the total expenditures were only $1.9 million. In contrast, the remaining medications in this category were those prescribed for relatively few patients at much higher costs. For example, there were 2567 claims for 205 enrollees ($15 970 per administration) for the prescription of recombinant antihemophilic factor VIII, with total expenditures of $42 million per year. These findings are consistent with previous research that revealed that 10 medication classes account for more than half of yearly Medicaid spending, with stimulant and antipsychotic prescriptions accounting for the highest aggregate expenditures.15 Our results also align by revealing that when medications are dispensed frequently, the aggregate expense is used to approximate the most expensive medications (eg, coagulation and anticoagulation factors); however, the comparative per-administration cost is considerably lower.

With our results, we highlight 2 primary mechanisms by which a child can accrue expenditures: (1) high use of low-cost services and (2) use of high-cost services. Therefore, for children with NC-CDs, we underscore the importance of considering a high use of low-cost services (eg, outpatient mental health services) in addition to the use of high-cost services (eg, high-cost medications or inpatient stays) when planning interventions. What remains unknown is whether the use of outpatient mental health services avoids a higher-cost inpatient stay or the use of higher-cost prescriptions. Because the outpatient provision of mental health services has been shown to reduce high resource use and subsequent outpatient costs,16,17 further investigation into this possibility may be warranted.

Our study has several limitations. The study population included only those who were enrolled in Medicaid and therefore may not be generalizable to all payer populations. However, children covered by commercial insurance have also demonstrated an increase in mental health service use since the Paul Wellstone and Pete Dominici Mental Health Parity and Addiction Equity Act of 2008.18 There is also a risk of misclassification of patient complexity when using the CRG algorithm. Children with NC-CDs often have relapsing and remitting courses and intermittent use of health care. Similarly, secondary to our use of claims data, children who have encounters during which a chronic condition is identified will be identified as such. Thus, children not specifically seeking care for chronic conditions may be undercounted. Therefore, the classification of NC-CDs by using CRGs likely lacks sensitivity.3 Misclassification would likely result in more children with NC-CDs and low health care spending. The spending patterns of the moderate and high expenditure groups reveal that spending patterns would remain similar; however, it is unclear how misclassification may influence the magnitude of our findings. In addition, adolescents with mental health conditions may not appropriately access needed mental health services.19 Additionally, the identification of mental health conditions by using Medicaid claims data likely underestimates the overall prevalence of mental health conditions for both inpatient and outpatient use.20 Also, our description of the overall economic burden for children with mental health conditions with Medicaid is likely underestimated. One reason may be that state-level health care policy may be used to reimburse mental health services outside of Medicaid, making our estimate artificially low.21 Omissions in ICD-9 coding for mental health conditions may lead to an underestimation of the economic effects of mental health conditions on increased health care use. We also included children without dominant chronic conditions from CRG 6, a group typically described as containing C-CDs. Our decision to include this sample of children in the NC-CD group was intended to improve clinical grouping and include children with 2 NC-CD conditions, who are placed in CRG 6 by the algorithm. Finally, because emergency department and inpatient expenditures for hospitalized patients are combined in the Truven MarketScan data set, expenditures for inpatient care may be overestimated by the cost of emergency department interventions. Similarly, outpatient visits that occur on the same day are grouped into 1 visit, which may underestimate outpatient use.

One percent of children in the ≥99th percentile of expenditures accounted for ∼20% of the annual expenditures for children with NC-CDs. Mental health conditions are associated with increases in health care expenditures, are common, and account for a large proportion of Medicaid spending in children with NC-CDs. Although outpatient mental health services account for substantial expenditures for children with NC-CDs, investment in these services may defray more expensive health care costs, such as inpatient admission.

     
  • CRG

    Clinical Risk Group

  •  
  • C-CD

    complex chronic disease

  •  
  • DRG

    Diagnosis-Related Group

  •  
  • ICD-9

    International Classification of Diseases, Ninth Revision

  •  
  • IQR

    interquartile range

  •  
  • NC-CD

    noncomplex chronic disease

  •  
  • PMPY

    per member per year

Dr Bettenhausen led the development of the study design, analysis, interpretation of the data, writing of the first manuscript draft, and integration of critical intellectual content during earlier versions of the manuscript; Drs Richardson, Shah, Hall, Andrews, Neff, Auger, Hoefgen, Macy, Plencner, and Zima participated in the study design, analysis, and interpretation of the data and provided critical intellectual content in the revision of the manuscript; and all authors approved the final manuscript as submitted.

FUNDING: No external funding.

1
Hoefgen
ER
,
Andrews
AL
,
Richardson
T
, et al
.
Health care expenditures and utilization for children with noncomplex chronic disease.
Pediatrics
.
2017
;
140
(
3
):
e20170492
2
Simon
TD
,
Cawthon
ML
,
Stanford
S
, et al;
Center of Excellence on Quality of Care Measures for Children with Complex Needs (COE4CCN) Medical Complexity Working Group
.
Pediatric medical complexity algorithm: a new method to stratify children by medical complexity.
Pediatrics
.
2014
;
133
(
6
). Available at: www.pediatrics.org/cgi/content/full/133/6/e1647
[PubMed]
3
Neff
JM
,
Clifton
H
,
Popalisky
J
,
Zhou
C
.
Stratification of children by medical complexity.
Acad Pediatr
.
2015
;
15
(
2
):
191
196
[PubMed]
4
Zima
BT
,
Rodean
J
,
Hall
M
,
Bardach
NS
,
Coker
TR
,
Berry
JG
.
Psychiatric disorders and trends in resource use in pediatric hospitals.
Pediatrics
.
2016
;
138
(
5
):
e20160909
[PubMed]
5
Doupnik
SK
,
Lawlor
J
,
Zima
BT
, et al
.
Mental health conditions and medical and surgical hospital utilization.
Pediatrics
.
2016
;
138
(
6
):
e20162416
[PubMed]
6
Bardach
NS
,
Coker
TR
,
Zima
BT
, et al
.
Common and costly hospitalizations for pediatric mental health disorders.
Pediatrics
.
2014
;
133
(
4
):
602
609
[PubMed]
7
Berry
JG
,
Hall
M
,
Neff
J
, et al
.
Children with medical complexity and Medicaid: spending and cost savings.
Health Aff (Millwood)
.
2014
;
33
(
12
):
2199
2206
[PubMed]
8
Centers for Medicare and Medicaid Services
. ICD-10-CM/PCS MS-DRG v34.0. Definitions manual. Available at: https://www.cms.gov/ICD10Manual/version34-fullcode-cms/fullcode_cms/P0001.html. Accessed July 24, 2018
9
Truven Health Analytics
. Micromedex Red Book. Available at: https://truvenhealth.com/products/micromedex/product-suites/clinical-knowledge/red-book. Accessed October 6, 2017
10
Kuo
DZ
,
Hall
M
,
Agrawal
R
, et al
.
Comparison of health care spending and utilization among children with Medicaid insurance.
Pediatrics
.
2015
;
136
(
6
). Available at: http://pediatrics.aappublications.org/content/136/6/1521
[PubMed]
11
Montalbano
A
,
Rodean
J
,
Kangas
J
,
Lee
B
,
Hall
M
.
Urgent care and emergency department visits in the pediatric Medicaid population.
Pediatrics
.
2016
;
137
(
4
):
e20153100
[PubMed]
12
Microsoft Azure
. K-means clustering. 2016. Available at: https://msdn.microsoft.com/en-us/library/azure/dn905944.aspx. Accessed May 24, 2017
13
Agrawal
R
,
Hall
M
,
Cohen
E
, et al
.
Trends in health care spending for children in Medicaid with high resource use.
Pediatrics
.
2016
;
138
(
4
):
e20160682
[PubMed]
14
Newman
L
,
Harris
V
,
Evans
LJ
,
Beck
A
.
Factors associated with length of stay in psychiatric inpatient services in London, UK.
Psychiatr Q
.
2018
;
89
(
1
):
33
43
[PubMed]
15
Cohen
E
,
Hall
M
,
Lopert
R
, et al
.
High-expenditure pharmaceutical use among children in Medicaid.
Pediatrics
.
2017
;
140
(
3
):
e20171095
[PubMed]
16
Malik
FS
,
Hall
M
,
Mangione-Smith
R
, et al
.
Patient characteristics associated with differences in admission frequency for diabetic ketoacidosis in United States children’s hospitals.
J Pediatr
.
2016
;
171
:
104
110
[PubMed]
17
Wright
DR
,
Katon
WJ
,
Ludman
E
, et al
.
Association of adolescent depressive symptoms with health care utilization and payer-incurred expenditures.
Acad Pediatr
.
2016
;
16
(
1
):
82
89
[PubMed]
18
Walter
AW
,
Yuan
Y
,
Cabral
HJ
.
Mental health services utilization and expenditures among children enrolled in employer-sponsored health plans.
Pediatrics
.
2017
;
139
(
suppl 2
):
S127
S135
[PubMed]
19
The Office of Adolescent Health, Department of Health and Human Services
. Adolescent access to mental health care. 2017. Available at: https://www.hhs.gov/ash/oah/adolescent-development/mental-health/access-adolescent-mental-health-care/index.html. Accessed March 31, 2017
20
National Institutes of Mental Health
. Prevalence of mental disorders among children. Available at: https://www.nimh.nih.gov/health/statistics/prevalence/any-disorder-among-children.shtml. Accessed March 31, 2017
21
Grimes
KE
,
Mullin
B
.
MHSPY: a children’s health initiative for maintaining at-risk youth in the community.
J Behav Health Serv Res
.
2006
;
33
(
2
):
196
212
[PubMed]

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data