Medicaid Expenditures on Psychotropic Medications for Maltreated Children: A Study of 36 States
Abstract
Objective:
Children with histories of abuse or neglect are the most expensive child population to insure for their mental health needs. This study aimed to quantify the magnitude of Medicaid expenditures incurred in the purchase of psychotropic drugs for these children.
Methods:
Children (N=4,445) participating in the National Survey of Child and Adolescent Well-Being (NSCAW) and from households under investigation for suspected child abuse and neglect were linked to their Medicaid claims from 36 states. Expenditures on psychotropic medications between the NSCAW sample and a propensity score–matched comparison sample of Medicaid-enrolled children were compared in a two-part regression of logistic and generalized linear models.
Results:
Children in the NSCAW sample had twice the odds of psychotropic drug use and $190 higher mean annual expenditures on psychotropic drugs than children in the comparison sample. Increased expenditures on antidepressants and antimanic drugs were the primary drivers of these increased expenditures. Male gender and white race-ethnicity were associated with significantly increased expenditures. Children in primary care case management had $325 lower expenditures than those in fee-for-service Medicaid. Among NSCAW children alone, male gender, older age, being in poorer health, and scoring in the clinical range of the Child Behavior Checklist (CBCL) all increased expenditures on psychotropic drugs.
Conclusions:
Medicaid agencies should focus their cost containment strategies on antidepressants and antimanic drugs, consider expanding primary care case management arrangements, and expand use of instruments such as the CBCL to identify and treat high-need children.
Children with histories of abuse and neglect—collectively termed “maltreatment”—are the predominant consumers of child mental health services in the United States today. In federal fiscal year 2011, approximately 6.2 million children nationwide were reported to child welfare or child protection agencies (referred to in this report as “child welfare”) for suspected maltreatment (1); half of all maltreated children have clinically significant emotional or behavioral problems (2,3). Medicaid is the dominant payer for the treatment of these problems (4), and quantifying such expenditures is critical to Medicaid agencies, especially as the Affordable Care Act exerts new pressures upon Medicaid budgets.
Approximately half of all spending for the treatment of emotional and behavioral disorders is on psychotropic drugs (5). Understanding this driver of mental health expenditures is critical for Medicaid agencies because children in child welfare receive psychotropic medications at a rate between two and three times that of comparable children in the community (6). Concomitant use of psychotropic drugs is more prevalent among children in child welfare, and they also receive more of these medications concomitantly (7,8); children in the child welfare system are now the largest consumers of psychotropic drugs among all child populations in the United States today.
The magnitude of such use has serious fiscal consequences for Medicaid agencies. Mental health costs for maltreated children can reach $16,848 per month per child (9), and total behavioral health expenditures incurred by each child in foster care are over eight times higher than expenditures incurred by nonfoster children (10). There is little information on the extent to which psychotropic medications are responsible for these expenditures, and extant studies examining this question have focused on either single-state analyses (11) or on smaller groupings of Medicaid states (12). Other child characteristics that might explain the need for such medications are also understudied at a population level. The fiscal utility of instruments purporting to capture mental health need, such as the Child Behavior Checklist (CBCL [13,14]), and maltreatment status is not well understood in large population-level data. This lack of information has been identified as one of the principal challenges facing Medicaid agencies in their attempts to contain costs of care for their child welfare beneficiaries (15).
Using a national panel survey of children and adolescents, who had contact with child welfare agencies, this study linked those data to Medicaid claims in 36 states. In this article, we quantify Medicaid expenditures on psychotropic medications among maltreated children and compare them with a propensity score–matched sample of child Medicaid beneficiaries without putative child welfare involvement. We modeled expenditures for both of these groups to identify drivers of expenditures, and we modeled expenditures in the sample of child welfare–involved children. Our aim was to provide information designed to help Medicaid agencies anticipate, better predict, and deliberately plan for mental health expenditures for their child welfare–involved beneficiaries.
Methods
Data sources and creation of analytic data set
The National Survey of Child and Adolescent Well-Being (NSCAW) was the first nationally representative panel study of children and adolescents coming into contact with child welfare agencies. The survey gathered data for 5,501 youths investigated by child protective services for possible abuse and neglect and 727 youths in long-term foster care placement in 92 primary sampling units in 97 counties throughout the United States. NSCAW’s baseline wave was conducted over a 15-month period beginning October 1999 (16,17), and these data were used for information on child and caregiver characteristics. We also obtained Medicaid claims files (Medicaid Analytic Extract, or “MAX” [18]) for years 2000 through 2003 and Medicaid enrollment files with Social Security Numbers (SSNs) and residence data of beneficiaries. We obtained data on all 36 states that were part of the NSCAW sampling frame.
We used SSNs to link 2,371 children who were NSCAW participants with their Medicaid records. For NSCAW children without an SSN match but for whom permission to match was available, we used all unique combinations of five-digit zip codes, date of birth, gender, and race-ethnicity to link these two data sets, for a linked sample of 4,445 children. (The remaining 1,783 children in NSCAW were not linkable either because their caregivers did not permit such linkage or because we lacked appropriate identifiers.)
We linked Medicaid enrollment files to drug claims files (RX file) across four years and aggregated individual claims within a single calendar year for a given NSCAW child. We deleted from our sample all children younger than two years at NSCAW wave 1 because NSCAW’s version of the CBCL (13,14) is not normed for that age group. Data for children who were not enrolled for at least ten months in a calendar year in select Medicaid plan types—fee-for-service (FFS), primary care case management (PCCM), or other managed care plans with non–mental health care carve-out—were also deleted, because we observed only enrollment, not claims or services, for children in Medicaid managed care. These steps left an analysis sample (nT) of 4,701 child-year observations, consisting of 1,861 unique children.
We generated a comparison sample of 4,701 child-year observations by using propensity score matching (19). We identified a cohort of Medicaid beneficiaries whose MAX records for all years indicated no eligibility for foster care; we then developed propensity scores by using age, gender, race-ethnicity, year of data, Medicaid plan type, and zip code of residence for all children. We then matched with replacement the NSCAW children to their nearest Medicaid neighbor. The resulting process yielded a sample that did not display statistically significant differences from NSCAW children with respect to age, gender, race-ethnicity, and plan type. Absence of foster care eligibility codes in Medicaid is an imperfect proxy for absence of maltreatment. It is likely that some of the children in the comparison group were subjected to maltreatment or had child welfare involvement that did not result in foster care placement; the magnitude of this bias is unknown.
Medicaid expenditures on psychotropic medications
We used codes from Medicaid RX (MRX) (20), the most widely used Medicaid pharmacy risk adjustment model, to aggregate drugs by indications of attention-deficit disorder, depression and anxiety, psychotic illness or bipolar disorder, and seizure disorder (to include anticonvulsants). Second, we used drug categories from the Red Book (21) in order to present information on drug classes of relevance to psychiatric practice. We purposefully selected mental health–relevant MRX and Red Book categories so that results from these two approaches would not be expected to be equivalent. Outcomes were measured and are reported as mean total annual Medicaid expenditure per child.
Covariates
Except where otherwise noted, all NSCAW variables were obtained from the child’s primary caregiver. Child-level covariates included child age, gender, and race-ethnicity. Identification of behavioral problems was based on whether the child scored in the clinical range (t score ≥64) on the internalizing or externalizing scales of the CBCL (13,14), a well-established measure of mental health need in child populations (6,22,23). Categories of physical abuse, sexual abuse, neglect, and abandonment were obtained from case workers and dichotomized such that a child could have more than one type of abuse coded. We also used a binary indicator variable representing fair or poor general physical health, with excellent, very good, or good as a referent, reported by the child’s caregiver.
Each child’s placement status was grouped into two mutually exclusive categories of in home (that is, living with his or her permanent primary caregiver, usually a birth parent) or out of home (in family foster care or in congregate care, such as a group home or residential treatment center). Information on whether the child lived in an urban or rural area was obtained from NSCAW data as a control for the availability of health care resources in the child’s community. We also included dummy variables for insurance type (FFS, PCCM, or both types) from the Medicaid enrollment files. All covariates were measured at baseline, except insurance type, which was measured at the child observation (a calendar year) level for each child.
Analyses
We first developed an aggregate expenditure figure per child per year for both NSCAW and comparison samples, and then we adjusted all expenditures to 2010 dollars (24,25). Bivariate analyses showing mean differences in rates of annual use of, and expenditures on, psychotropic medications between NSCAW and comparison group children were performed with two-sample proportions and t tests.
Differences in psychotropic medication expenditures between children in NSCAW and the comparison group were examined with a two-part (26) model; expenditure per child per year was its outcome. In the first part we used logistic regression to estimate the annual probability of having any medication expenditures, and in the second part we used a generalized linear model (GLM) with a log link and a gamma distribution (27,28). We estimated similar models with an NSCAW-only sample (nT=3,520 child-year observations after accounting for some missing values and without the propensity score–matched sample) to examine the association between NSCAW’s rich set of explanatory variables and expenditures. We present the combined marginal effect showing the joint impact on Medicaid expenditures of both differences in use (part 1) and levels of expenditure (part 2).
We report unweighted expenditure data in keeping with prior literature (12,29). All models included corrections for the clustering of multiple years’ worth of expenditure observations per child. We also included state dummies to control for unobserved state-level variables and year dummies to control for secular trends (not shown in tables).
All analyses were performed in Stata, version 13.1 (30). NSCAW participants gave informed consent to these data linkages, and these analyses were approved by the Washington University Human Research Protection Office and the Institutional Review Board of Research Triangle Institute (RTI International).
Results
Among the observations of NSCAW youths (measured at the child-year level), a total of 2,289 (48.7%) were male. At NSCAW’s baseline wave, 1,634 (34.8%) were between ages two and five years, 1,666 (35.4%) were between ages six and 11, 521 (11.1%) were ages 12 or 13, and 880 (18.7%) were age 14 or older. Most children (N=2,480, 52.8%) were of non-Hispanic white race-ethnicity, others were black (N=1,532, 32.6%), Hispanic (N=411, 8.8%), or of other race-ethnicity (N=81, 1.7%), and the remainder were of unknown race-ethnicity (N=197, 4.2%).
Table 1 shows bivariate analyses of the mean differences in utilization of, and annual expenditures on, psychotropic medications between youths in the NSCAW sample and the comparison group. Prescriptions in the MRX classifications shown were used by 25% of the NSCAW sample, versus by 16% of the comparison sample (p<.001). Among those who received these medications, mean drug expenditures for NSCAW youths were significantly higher ($1,559) than those of children in the comparison group ($1,300; p<.001). NSCAW children had significantly higher use of and expenditures overall for drugs specified in the MRX system. Results of differences in pharmaceutical class (Red Book classifications) were similar in direction and significance. Compared with the control group, NSCAW participants had significantly higher expenditures on antimanic agents ($143 higher) and antidepressants ($82 higher) and significantly lower expenditures on benzodiazepine-type sedatives or hypnotics ($158 lower).
| Comparison sample | NSCAW sampleb | |||||||
|---|---|---|---|---|---|---|---|---|
| Utilization | Expenditure | Utilization | Expenditure | p | ||||
| Drug | N | % | ($) | N | % | ($) | Utilization | Expenditure |
| MRX classification | ||||||||
| Attention-deficit hyperactivity disorder | 471 | 10.0 | 561 | 719 | 15.3 | 594 | <.001 | ns |
| Depression or anxiety disorder | 360 | 7.7 | 512 | 593 | 12.6 | 599 | <.001 | <.01 |
| Psychotic illness or bipolar disorder | 200 | 4.3 | 1,724 | 391 | 8.3 | 1,797 | <.001 | ns |
| Seizure disorder | 185 | 3.9 | 1,003 | 327 | 7.0 | 954 | <.001 | ns |
| Total | 753 | 16.0 | 1,300 | 1,152 | 24.5 | 1,559 | <.001 | <.001 |
| Red Book classification | ||||||||
| Sedatives or hypnotics, barbiturates | 21 | .4 | 77 | 9 | .2 | 117 | <.01 | ns |
| Sedatives or hypnotics, benzodiazepines | 34 | .7 | 220 | 29 | .6 | 62 | ns | <.01 |
| Anticholinergic, antimuscarinic, and antispasmodic | 53 | 1.1 | 73 | 39 | .8 | 46 | ns | ns |
| Anticholinergic or antiparkinsonism agent | 10 | .2 | 62 | 28 | .6 | 74 | <.001 | ns |
| Anticholinergic, not elsewhere classified | 12 | .3 | 261 | 14 | .3 | 167 | ns | ns |
| Anticonvulsants, hydantoin derivative | 17 | .4 | 238 | 5 | .1 | 143 | <.01 | ns |
| Anticonvulsants, oxazolidinediones | 0 | — | — | 0 | — | — | ns | |
| Anticonvulsants, succinimides | 1 | — | — | 0 | — | — | ns | |
| Anticonvulsants, benzodiazepine | 13 | .3 | 156 | 23 | .5 | 161 | <.05 | ns |
| Anticonvulsants, miscellaneous | 168 | 3.6 | 1,047 | 319 | 6.8 | 970 | <.001 | ns |
| Antimanic agents, not elsewhere classified | 18 | .4 | 112 | 44 | .9 | 255 | <.001 | <.01 |
| Anxiolytics, sedatives, or hypnotics not elsewhere classified | 168 | 3.6 | 100 | 185 | 3.9 | 91 | ns | ns |
| Central nervous system agents, miscellaneous | 39 | .8 | 440 | 60 | 1.3 | 499 | <.01 | ns |
| Opiate antagonists, not elsewhere classified | 1 | — | — | 1 | — | — | ns | |
| Antidepressants | 323 | 6.9 | 519 | 564 | 12.0 | 601 | <.001 | <.01 |
| Antipsychotics | 197 | 4.2 | 1,743 | 382 | 8.1 | 1,810 | <.001 | ns |
| Stimulant, amphetamine type | 458 | 9.7 | 540 | 704 | 15.0 | 565 | <.001 | ns |
| Stimulant, nonamphetamine | 3 | .1 | — | 0 | — | — | <.05 | |
| Total | 909 | 19.3 | 1,097 | 1,277 | 27.2 | 1,417 | <.001 | <.001 |
a
nT=9,402 child-year observations. Utilization is the row mean and is nonexclusive, so the sum may exceed the Medicaid drug claims file (MRX) or Red Book total. Expenditures reflect means for observations of children using a particular type of medication. Zeroes are not included in the means.
b
NSCAW, National Survey of Child and Adolescent Well-Being. Participants are children and adolescents who have come into contact with child welfare agencies.
Differences in cumulative drug expenditures between NSCAW and comparison groups are reported in Table 2. Odds ratios (ORs) from part 1 of the model indicate the odds of incurring any expenditures on psychotropic drugs, according to the MRX classification; for brevity, we do not show expenditure differences for the Red Book classifications. An NSCAW child had nearly twice the adjusted odds of incurring an expenditure on a psychotropic drug compared with a non-NSCAW child (OR=1.93, p<.001). Across both groups, males and older children had significantly higher odds of incurring any expenditures, whereas children who were black or Hispanic had lower odds than white children of incurring any expenditures. Children of other or unknown races-ethnicities had significantly higher odds of medication use, but their small numbers made interpretation of these ORs problematic. Compared with children in FFS Medicaid plans, children in PCCM Medicaid plans had about 50% lower odds of incurring any psychotropic medication expenditure (OR=.48; p<.001).
| Part 1: any expenditure | Part 2: expenditure if >$0 | Combined marginal effect (parts 1 and 2)b | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Characteristic | OR | 95% CI | p | GLM coeffc | 95% CI | p | $ | 95% CI | p |
| NSCAW sample (reference: comparison sample) | 1.93 | 1.67 to 2.24 | <.001 | .25 | .12 to .39 | <.001 | 189.83 | 136.32 to 243.34 | <.001 |
| Male (reference: female) | 1.90 | 1.63 to 2.22 | <.001 | .15 | .001 to .30 | ≤.05 | 200.32 | 129.37 to 271.27 | <.001 |
| Age (reference: 3–5) | |||||||||
| 6–11 | 9.03 | 7.12 to 11.47 | <.001 | .62 | .33 to .92 | <.001 | 894.86 | 602.97 to 1,186.76 | <.001 |
| 12–13 | 12.19 | 9.31 to 15.97 | <.001 | .97 | .66 to 1.28 | <.01 | 1,790.35 | 1,165.39 to 2,415.30 | <.001 |
| ≥14 | 13.89 | 10.65 to 18.12 | <.001 | 1.11 | .81 to 1.42 | <.001 | 1,801.08 | 1,261.39 to 2,340.76 | <.001 |
| Race-ethnicity (reference: white) | |||||||||
| Black | .68 | .56 to .82 | <.001 | –.20 | –.38 to –.02 | ≤.05 | –153.80 | –223.86 to –83.74 | <.001 |
| Hispanic | .68 | .48 to .95 | ≤.05 | –.10 | –.39 to .19 | ns | –112.84 | –221.30 to –4.38 | ≤.05 |
| Other or unknown | 1.59 | 1.15 to 2.20 | <.01 | .39 | .11 to .68 | <.01 | 331.94 | 91.33 to 572.55 | <.01 |
| Insurance (reference: fee for service) | |||||||||
| Other and multiple types | 1.17 | .72 to 1.90 | ns | .23 | –.41 to .87 | ns | 139.70 | –223.07 to 502.48 | ns |
| PCCM onlyd | .48 | .39 to .61 | <.001 | –.60 | –.81 to –.39 | <.001 | –325.06 | –409.83 to –240.28 | <.001 |
a
NSCAW, National Survey of Child and Adolescent Well-Being. nT=9,402 child-year observations. All models included state and year dummy variables (not shown) to control for state Medicaid differences and time trends. All insurance categories in the model included full behavioral health coverage.
b
Reflects both differences in the likelihood of use of (part 1) and levels of expenditure on (part 2) psychotropic drugs
c
Generalized linear model coefficient
d
Primary care case management
Table 2 also displays coefficients of the GLM (part 2), which indicates predictors of expenditure among children with any (nonzero) expenditures on psychotropic drugs. The direction of these predictors largely paralleled the ORs from part 1 of the model. Among only those with positive expenditure in these MRX categories, children who had participated in the NSCAW incurred an average of $190 more in annual psychotropic drug expenditures compared with children in the control group. The purchase of psychotropic drugs for a male child cost Medicaid $200 more on average annually than for a female child, and $1,801 more for a child age 14 or older. Black children and Hispanic children incurred $154 and $113 less mean annual expenditures compared with white children. A child whose Medicaid program reimbursed providers on a PCCM basis incurred $325 less in expenditures than a child whose Medicaid program paid its providers via FFS.
Table 3 displays results from a two-part model of differences in psychotropic medication expenditures conducted on a stratified sample of the children in the NSCAW group. Many of the demographic findings are similar to those shown in Table 2. An important finding from this model is that children placed in foster care or in residential care incurred an average expenditure of $168 more each year on psychotropic drugs. Children in poor or fair health incurred $228 more in expenditures compared with children in excellent or good health. CBCL scores were good predictors of expenditures; a clinically significant externalizing CBCL score was associated with $555 more, whereas a clinically significant internalizing score was associated with $152 more, in annual psychotropic drug expenditures per child. Maltreatment history did not seem to be an independent risk factor for psychotropic drug expenditures.
| Part 1: any expenditure | Part 2: expenditure if >$0 | Combined marginal effect (parts 1 and 2)b | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Characteristic | OR | 95% CI | p | GLM coeffc | 95% CI | p | $ | 95% CI | p |
| Male (reference: female) | 2.11 | 1.76 to 2.52 | <.001 | .20 | .04 to .36 | <.05 | 252.88 | 135.71 to 370.06 | <.001 |
| Age (reference: 3–5) | |||||||||
| 6–11 | 6.27 | 4.31 to 9.12 | <.001 | .19 | –.20 to .58 | ns | 441.49 | 147.86 to 735.12 | <.01 |
| 12–13 | 6.54 | 4.32 to 9.90 | <.001 | .36 | –.05 to .78 | ns | 676.93 | 175.47 to 1,178.39 | <.01 |
| ≥14 | 6.88 | 4.62 to 10.24 | <.001 | .46 | .05 to .87 | <.05 | 723.44 | 288.76 to 1,158.12 | <.01 |
| Race-ethnicity (reference: white) | |||||||||
| Black | .65 | .53 to .81 | <.001 | –.38 | –.57 to –.19 | <.001 | –251.89 | –370.88 to –132.90 | <.01 |
| Hispanic | .92 | .67 to 1.25 | ns | –.01 | –.28 to .27 | ns | –20.78 | –215.55 to 173.99 | ns |
| Other or unknown | 1.03 | .73 to 1.46 | ns | –.20 | –.51 to .10 | ns | –86.43 | –285.48 to 112.63 | ns |
| Insurance (reference: fee for service) | |||||||||
| Other and multiple insurance types | .91 | .46 to 1.81 | ns | .82 | .08 to 1.55 | <.05 | 532.34 | –99.39 to 1,164.08 | ns |
| PCCM onlyd | .62 | .47 to .82 | <.01 | –.67 | –.91 to –.43 | <.001 | –355.36 | –472.39 to –238.33 | <.001 |
| Out-of-home care (reference: in-home care) | 2.03 | 1.67 to 2.47 | <.001 | .04 | –.12 to .21 | ns | 168.47 | 46.55 to 290.38 | <.01 |
| Fair or poor general physical health (reference: excellent, very good, or good health) | 1.97 | 1.10 to 3.35 | <.001 | .14 | –.07 to .35 | ns | 228.25 | 12.87 to 443.63 | <.05 |
| Rural (reference: urban) | 1.08 | .39 to 2.33 | ns | –.03 | –.24 to .19 | ns | 1.57 | –153.14 to 156.28 | ns |
| Child Behavior Checklist (reference: t score <64) | |||||||||
| Externalizing t score ≥64 | 4.43 | 3.65 to 5.37 | <.001 | .49 | .32 to .67 | <.001 | 555.39 | 425.04 to 685.73 | <.001 |
| Internalizing t score ≥64 | 1.38 | 1.13 to 1.68 | <.01 | .17 | .004 to .34 | <.05 | 152.31 | 23.35 to 281.28 | <.05 |
| Maltreatment history (reference: none) | |||||||||
| Physical abuse | 1.41 | 1.14 to 1.73 | <.01 | –.03 | –.21 to .15 | ns | 55.85 | –79.93 to 191.63 | ns |
| Sexual abuse | 1.30 | 1.00 to 1.67 | <.05 | .22 | .01 to .44 | <.05 | 179.34 | –21.80 to 380.47 | ns |
| Neglect | .82 | .66 to 1.01 | ns | –.01 | –.20 to .18 | ns | –47.42 | –187.53 to 92.69 | ns |
| Abandonment | 1.30 | .90 to 1.89 | ns | .50 | .17 to .82 | <.01 | 385.28 | –7.70 to 778.25 | ns |
a
nT=3,520 child-year observations. All models include state and year dummy variables (not shown) to control for state Medicaid differences and time trends. All insurance categories in the model include full behavioral health coverage.
b
Reflects both differences in the likelihood of use of (part 1) and levels of expenditure on (part 2) psychotropic drugs
c
Generalized linear model coefficient
d
Primary care case management
Discussion
In this study we examined in 36 states Medicaid expenditures on psychotropic drugs among a sample of youths who were respondents to the NSCAW. These children had almost twice the odds of using medications and incurred between 20% and 30% higher expenditures on medications compared with Medicaid child enrollees without apparent foster care involvement. Each maltreated child enrolled into Medicaid increased the program’s expenditures on psychotropic medications by approximately $190 in mean annual psychotropic drug expenditures among children with nonzero expenditures. These estimates, spread over 36 state Medicaid programs, provide greater precision than prior attempts to identify the magnitude of incurred expenditures with this population.
These mean annual expenditures also need to be considered longitudinally. Median length of stay in foster care for children who were finally adopted was nearly three years (31). Even after departure from foster care, children maintain Medicaid eligibility for a mean of three months (32). Consequently, cumulative median Medicaid expenditures on medications can approximate $600 at a lower bound, with a large proportion of children—especially older children who leave foster care when they attain the age of legal adulthood—costing Medicaid agencies several thousand dollars throughout their stay in the child welfare system.
Our findings are conservative compared with prior research suggesting that maltreated children increase Medicaid expenditures by between $237 and $840 per year (12). One reason for this difference is that our sample was much larger than the sample for the previous study, with greater precision in its estimates. But, more important, prior work on expenditures has used conditional marginal effects (examining only those with nonzero expenditures) to arrive at expenditure estimates. This article presents the joint marginal effect (which includes information from both parts of the two-part model). Hence, even though the expenditure difference looks smaller, it may be the more relevant number for Medicaid programs because the full budgetary impact of child maltreatment is felt through both differences in use (part 1) as well as through expenditures conditional on use (part 2).
The implications for Medicaid cost-containment policies were also clearer in this study. The use of antimanic and antidepressant medications seems to drive much of the expenditure differences between NSCAW and comparison groups. Focusing quality improvement and prior-authorization programs (33) on these two drug classes may be worthwhile. It is also intriguing that PCCM plans, compared with traditional FFS plans, produced mean annual cost savings of $355 on psychotropic medications. Because our focus was on costs, and not on quality or outcomes, we cannot comment on the appropriateness of such a structural arrangement. However, if these savings were due to better care coordination rather than increased unmet need, then this may support the efforts of Medicaid agencies to move child welfare–involved children into medical-home models.
Finally, the CBCL remains a powerful predictive instrument to estimate costs of psychotropic medications. Clinical scores on the externalizing and internalizing subscales were associated with an increase of $555 and $152 in mean annual expenditures for psychotropic drugs. The CBCL is highly rated in the child welfare field (34), and our findings offer Medicaid agencies a financial reason for its adoption as a population-level screening instrument.
Our study was subject to a few limitations. The design of our data linkage and our inability to use weights means that our data were convenience samples of children in 36 states. Second, we used Medicaid eligibility codes to identify a comparison sample of child Medicaid beneficiaries without foster care involvement. It is possible that some of these children were maltreated, in which case our estimates of expenditure differences between NSCAW children and comparison children are conservatively biased. Third, our data are reflective only of children in nonmanaged Medicaid systems, which form the largest type of payment systems in child welfare (4), and were the dominant plan types for child welfare participants in our sample.
Conclusions
Despite these limitations, this linkage between survey data and Medicaid claims data in 36 states provides information to Medicaid policy makers for better predicting psychotropic medication expenditures for a highly vulnerable population. Planning for these expenditures and ensuring that the needs of the most emotionally disturbed children are adequately resourced are critically important to Medicaid agencies as they attempt to resource care for children in the child welfare system within an increasingly unstable and uncertain fiscal climate.
Acknowledgments and disclosures
This study was funded by grants R01 MH092312 and T32MH019960 from the National Institute of Mental Health (NIMH), grant R01 HS020269 from the Agency for Healthcare Research and Quality, and grant HHSN271201200644P from NIMH’s Office for Research in Disparities and Global Mental Health. The NSCAW was developed under contract with the Administration on Children, Youth, and Families, U.S. Department of Health and Human Services (ACYF/DHHS). The data were provided by the National Data Archive on Child Abuse and Neglect. The information and opinions expressed in this article reflect solely the position of the authors. Nothing herein should be construed to indicate the support or endorsement of its content by ACYF/DHHS, the Centers for Disease Control and Prevention, NIMH, or the National Institutes of Health.
The authors report no competing interests.
References
1.
Child Maltreatment 2011. Washington, DC, Children’s Bureau, US Department of Health and Human Services. Available at www.acf.hhs.gov/programs/cb/resource/child-maltreatment-2011. Accessed May 13, 2013
2.
Burns BJ, Phillips SD, Wagner HR, et al: Mental health need and access to mental health services by youths involved with child welfare: a national survey. Journal of the American Academy of Child and Adolescent Psychiatry 43:960–970, 2004
3.
Hurlburt MS, Leslie LK, Landsverk J, et al: Contextual predictors of mental health service use among children open to child welfare. Archives of General Psychiatry 61:1217–1224, 2004
4.
Raghavan R, Leibowitz AA: Medicaid and mental health care for children in the child welfare system; in Child Protection: Using Research to Improve Policy and Practice. Edited by Haskins R, Wulczyn F, Webb MB. Washington, DC, Brookings Institute, 2007
5.
Frank RG, Goldman HH, McGuire TG: Trends in mental health cost growth: an expanded role for management? Health Affairs 28:649–659, 2009
6.
Raghavan R, Zima BT, Andersen RM, et al: Psychotropic medication use in a national probability sample of children in the child welfare system. Journal of Child and Adolescent Psychopharmacology 15:97–106, 2005
7.
Zito JM, Safer DJ, Sai D, et al: Psychotropic medication patterns among youth in foster care. Pediatrics 121:e157–e163, 2008
8.
Raghavan R, McMillen JC: Use of multiple psychotropic medications among adolescents aging out of foster care. Psychiatric Services 59:1052–1055, 2008
9.
Conrad C: Measuring costs of child abuse and neglect: a mathematic model of specific cost estimations. Journal of Health and Human Services Administration 29:103–123, 2006
10.
Becker M, Jordan N, Larsen R: Behavioral health service use and costs among children in foster care. Child Welfare 85:633–647, 2006
11.
Dosreis S, Yoon Y, Rubin DM, et al: Antipsychotic treatment among youth in foster care. Pediatrics 128:e1459–e1466, 2011
12.
Raghavan R, Brown DS, Thompson H, et al: Medicaid expenditures on psychotropic medications for children in the child welfare system. Journal of Child and Adolescent Psychopharmacology 22:182–189, 2012
13.
Achenbach TM: Manual for the Child Behavior Checklist/4–18 and 1991 Profile. Burlington, University of Vermont, Department of Psychiatry, 1991
14.
Achenbach TM: Manual for the Child Behavior Checklist/2–3 and 1992 Profile. Burlington, University of Vermont, Department of Psychiatry, 1992
15.
Raghavan R: Using risk adjustment approaches in child welfare performance measurement: applications and insights from health and mental health settings. Children and Youth Services Review 32:103–112, 2010
16.
Dowd K, Kinsey S, Wheeless A, et al: National Survey of Child and Adolescent Well-Being: Combined Waves 1–4, Data File User’s Manual. Ithaca, NY, National Data Archive on Child Abuse and Neglect, 2006
17.
NSCAW Research Group: Methodological lessons from the National Survey of Child and Adolescent Well-Being: the first three years of the USA’s first national probability study of children and families investigated for abuse and neglect. Children and Youth Services Review 24:513–541, 2002
18.
MAX 1999 and later. Minneapolis, Minn, Research Data Assistance Center.Available at www.resdac.org/Medicaid/max-1999.asp. Accessed Jan 20, 2010
19.
Guo S, Fraser MW: Propensity Score Analysis: Statistical Methods and Applications. Los Angeles, Sage, 2010
20.
Gilmer T, Kronick R, Fishman P, et al: The Medicaid Rx model: pharmacy-based risk adjustment for public programs. Medical Care 39:1188–1202, 2001
21.
Red Book. Bethesda, Md, Truven Health Analytics. Available at sites.truvenhealth.com/redbook/index.html. Accessed Feb 11, 2014
22.
Raghavan R, Inoue M, Ettner SL, et al: A preliminary analysis of the receipt of mental health services consistent with national standards among children in the child welfare system. American Journal of Public Health 100:742–749, 2010
23.
Raghavan R, Lama G, Kohl P, et al: Interstate variations in psychotropic medication use among a national sample of children in the child welfare system. Child Maltreatment 15:121–131, 2010
24.
Using Appropriate Price Indices for Expenditure Comparisons. Bethesda, Md, US Department of Health and Human Services, Agency for Healthcare Research and Quality. Available at www.meps.ahrq.gov/mepsweb/about_meps/Price_Index.shtml. Accessed March 5, 2010
25.
National Health Expenditures Accounts: Definitions, Sources, and Methods. Baltimore, Centers for Medicare and Medicaid Services Office of the Actuary, 2008. Available at www.cms.hhs.gov/NationalHealthExpendData/downloads/dsm-08.pdf. Accessed March 5, 2010
26.
Duan N, Manning WG, Morris CN, et al: Choosing between the sample-selection model and the multi-part model. Journal of Business and Economic Statistics 2:283–289, 1984
27.
Manning WG, Basu A, Mullahy J: Generalized modeling approaches to risk adjustment of skewed outcomes data. Journal of Health Economics 24:465–488, 2005
28.
Manning WG, Basu A, Mullahy J: Generalized Modeling Approaches to Risk Adjustment of Skewed Outcomes Data. Technical Working Paper 293. Washington DC, National Bureau of Economic Research, 2003. Available at www.nber.org/papers/t0293.pdf
29.
Florence C, Brown DS, Fang X, et al: Health care costs associated with child maltreatment: impact on Medicaid. Pediatrics 132:312–318, 2013
30.
Stata Statistical Software: Release 13.1. College Station, Tex, Stata Corp, 2013
31.
Wildfire J, Barth RP, Green RL: Predictors of reunification; in Child Protection: Using Research to Improve Policy and Practice. Edited by Haskins R, Wulczyn F, Webb MB. Washington, DC, Brookings Institution Press, 2007
32.
Raghavan R, Shi P, Aarons GA, et al: Health insurance discontinuities among adolescents leaving foster care. Journal of Adolescent Health 44:41–47, 2009
33.
Tilly J, Elam L: Prior Authorization for Medicaid Prescription Drugs. Washington, DC, Kaiser Family Foundation, 2003
34.
Child Behavior Checklist for Ages 6–18 (CBCL/6-18). California Evidence-Based Clearinghouse for Child Welfare, 2010. Available at www.cebc4cw.org/assessment-tool/child-behavior-checklist-for-ages-6-18/. Accessed April 3, 2011
Information & Authors
Information
Published In
Cover: Two Children Watching the Clouds in a Field, by Elizabeth Shippen Green. © Copyright 2014 National Museum of American Illustration™, Newport, RI. Photos courtesy Archives of the American Illustrators Gallery™, New York City.
History
Published ahead of print: 17 November 2014
Published online: 1 December 2014
Published in print: December 01, 2014
Authors
Funding Information
Agency for Healthcare Research and Quality10.13039/100000133: R01 HS020269
National Institute of Mental Health10.13039/100000025: R01 MH092312, 2T32 HL007456-26, T32MH019960, HHSN271201200644P
Metrics & Citations
Metrics
Citations
Export Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
For more information or tips please see 'Downloading to a citation manager' in the Help menu.
View Options
View options
PDF/EPUB
View PDF/EPUBLogin options
Already a subscriber? Access your subscription through your login credentials or your institution for full access to this article.
Personal login Institutional Login Open Athens loginNot a subscriber?
PsychiatryOnline subscription options offer access to the DSM-5-TR® library, books, journals, CME, and patient resources. This all-in-one virtual library provides psychiatrists and mental health professionals with key resources for diagnosis, treatment, research, and professional development.
Need more help? PsychiatryOnline Customer Service may be reached by emailing [email protected] or by calling 800-368-5777 (in the U.S.) or 703-907-7322 (outside the U.S.).