0
We're unable to sign you in at this time. Please try again in a few minutes.
Retry
We were able to sign you in, but your subscription(s) could not be found. Please try again in a few minutes.
Retry
There may be a problem with your account. Please contact the AMA Service Center to resolve this issue.
Contact the AMA Service Center:
Telephone: 1 (800) 262-2350 or 1 (312) 670-7827  *   Email: subscriptions@jamanetwork.com
Error Message ......
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 50.17.176.149. Please contact the publisher to request reinstatement.
Original Article |

The Effect of a Managed Behavioral Health Carve-Out on Quality of Carefor Medicaid Patients Diagnosed as Having Schizophrenia FREE

Alisa B. Busch, MD, MS; Richard G. Frank, PhD; Anthony F. Lehman, MD, MSPH
[+] Author Affiliations

From the Departments of Psychiatry (Dr Busch) and Health Care Policy(Dr Frank), Harvard Medical School, Boston, Mass; Department of Psychiatry,University of Maryland School of Medicine, Baltimore, Md (Dr Lehman); Alcoholand Drug Abuse Treatment Program, McLean Hospital, Belmont, Mass (Dr Busch);and National Bureau of Economic Research, Cambridge, Mass (Dr Frank).


Arch Gen Psychiatry. 2004;61(5):442-448. doi:10.1001/archpsyc.61.5.442.
Text Size: A A A
Published online

Context  Managed behavioral health carve-outs (MBHCOs) are a regular feature of public and private mental health care systems and have been successful in reducing costs. The evidence on quality impacts is limited and suggests comparable quality overall, except that people with severe psychiatric disorders may be those most disadvantaged by MBHCOs.

Objective  To explore the effect of implementing an MBHCO on the quality of outpatient care received by enrollees diagnosed as having schizophrenia.

Design and Participants  Observational retrospective cohort study using a quasi-experimental design of state Medicaid enrollees diagnosed as having schizophrenia, aged 18 to 64 years between 1994 and 2000 in the carve-out and comparison regions (8082 person-years).

Setting  Ambulatory care.

Main Outcome Measures  Quality indicators derived from the Schizophrenia Patient Outcomes Research Team recommendations.

Results  There was no statistical difference between the carve-out and integrated arrangements in the likelihood of receiving any antipsychotic medication (odds ratio [OR], 1.02; 95% confidence interval [CI], 0.81-1.29), second-generation antipsychotics (including clozapine: OR, 1.05; 95% CI, 0.86-1.28; not including clozapine: OR, 1.05; 95% CI, 0.85-1.29), or antiextrapyramidal medication (OR, 1.36; 95% CI, 0.84-2.19). The carve-out was negatively associated with receiving any individual therapy (OR, 0.27; 95% CI, 0.22-0.33), group therapy (OR, 0.19; 95% CI, 0.14-0.25), and psychosocial rehabilitation (OR, 0.31; 95% CI, 0.26-0.38). Family therapy occurred for less than 1% of this population in both carve-out and integrated regions.

Conclusions  The MBHCO was not associated with changes in medication quality (for which it was not at financial risk). It was significantly associated with sharp decreases in the likelihood of receiving psychosocial treatments (for which it was financially at risk)—independent of whether a clinical evidence base supported them.

Managed behavioral health carve-outs (MBHCOs) are a regular featureof the mental health care delivery landscape1 andhave been successful in reducing costs.24 Asstates struggle to control Medicaid spending, they have turned to managedbehavioral health care organizations to ration public mental health–substanceabuse (MH/SA) care. More than 20 states have adopted carve-outs to manageMH/SA care in the Medicaid program.1 Carve-outsserve to separate part of the risk in a health insurance arrangement and manageit under a separate contract. Two sets of health insurance risks where thisoccurs frequently are mental health and substance abuse care (together referredto as behavioral health care) and prescription medications.There are several different approaches to organizing carve-out arrangements.5

There are multiple ways in which states can contract with carve-outs.In the analysis presented in this article, we focus on a carve-out contractwhere the state directly contracts with a specialty managed behavioral healthcare vendor. This arrangement is quite common within state Medicaid programs.Arizona, Colorado, Florida, Iowa, Massachusetts, Maryland, New Jersey, andUtah, among others, have adopted this approach. Under such contracts, specialtymental health services—which include psychotherapy, medication management,day treatment, inpatient care, psychosocial rehabilitation, and case managementservices—are managed by the carve-out vendor. The contract frequentlyincludes financial risk for the carve-out vendor and some performance requirements.Risk-based contracts range from pure capitation arrangements to contractsspecifying high levels of risk sharing between the state Medicaid programand the vendor. Performance requirements span from the speed at which telephonesfor intake are answered to indicators of continuity of care. Carve-out contractsgenerally do not include management of prescription drug utilization.

The MBHCOs offer potential advantages and disadvantages. Among the disadvantagesare incentives for cost shifting and concerns regarding barriers to accessand coordination of care.5 The advantages includethe application of specialized expertise to the rationing of MH/SA care, scaleeconomies for smaller health plans, and, in some circumstances, protectionagainst adverse selection.5,6 Theidea of using clinical expertise, instead of cost-sharing provisions and limitson service use, to ration care offers the potential to contain costs and tomaintain or improve quality of care by precisely targeting waste and inappropriatetreatment. Additionally, such expertise can be used to develop performancestandards and monitor the quality of care that patients receive.7 Existingresearch suggests that there are abundant opportunities to eliminate inefficientand ineffective MH/SA care.8 Extant evidenceindicates that Medicaid MBHCOs decrease costs in the specialty mental healthsector,2,3,6,9 butthe literature is mixed with respect to the effect of MBHCO arrangements onquality of care, particularly for severely and persistently mentally ill patientsin Medicaid.13,815

This study uses a natural experiment in implementing an MBHCO withina state Medicaid program. Because existing evidence points to adverse impactsfor people with severe mental disorders under managed care arrangements, wefocus on the quality of care provided to people diagnosed as having schizophrenia.We make use of the treatment recommendations from the Schizophrenia PatientOutcomes Research Team (PORT)16 to measurequality of care. In addition to quality measures recommended by the PORT,we also include one measure (ie, psychosocial rehabilitation) that the PORTdid not endorse because of an inadequate evidence base. This was to answerthe following question: Does an MBHCO that is meant to apply specialized expertisein its care management discriminate between treatments supported by evidence-basedrecommendations and those that are not?

THE EXPERIMENTAL CONTEXT

Before the implementation of the MBHCO in 1996, the state's Medicaidenrollees were served in a fee-for-service program where the primary carephysicians also received a capitation payment for providing gatekeeping andcase management services. The exception was that persons enrolled in a stateMedicaid health maintenance organization (HMO) had their mental health servicesmanaged by the HMO. In 1996, the state obtained a 1915b waiver from the federalgovernment to implement a prepaid mental health plan demonstration. As a result,a private for-profit MBHCO vendor was awarded a contract to manage specialtyMH/SA Medicaid services in one region of the state. Enrollees in HMOs wereexcluded from the carve-out arrangement. The contract was a "full-risk" orcapitation contract, meaning that the MBHCO carried 100% of the financialrisk for mental health costs (inpatient and outpatient treatments), excludingprescription drug costs. This means that prescription costs were borne bythe state, not the carve-out. The carve-out, in turn, shared financial riskwith local community mental health centers (CMHCs). The CMHCs received capitationpayments, but 4% was held back and put into a risk pool to cover cost overruns.The CMHCs were able to recoup 50 cents on the dollar in excess costs up totheir total contribution to the risk pool; further costs were reimbursed at25 cents on the dollar. Thus, the CMHCs were buffered from the full loss ifthey exceeded the capitation amount. Since CMHCs were responsible for costsincurred, they were also responsible for utilization management. The carve-outand CMHCs developed guidelines for medical necessity, length of stay, anddiagnosis-based treatment protocols. Thus, CMHCs did not seek "authorization"but rather notified the carve-out concerning services required for an individualpatient. Services delivered that were not in concert with the establishedprotocols did not count toward compensation from the risk pool. The remainderof the state's non-HMO Medicaid program remained in the lightly managed fee-for-servicesystem with primary care gatekeepers. This created a natural experiment ina carve-out arrangement.

STUDY DESIGN AND SOURCES OF DATA

The cohort included Medicaid enrollees from July 1, 1994, through June30, 2000. The structure of this natural experiment allowed us to implementa quasi-experimental design. That is, the region where the carve-out was introducedwas viewed as the experimental intervention and 2 similarly urban regionswere chosen as controls. The data were obtained from the state and includedadministrative records on service utilization and spending. The administrativeclaims files contained records of inpatient hospitalizations, outpatient treatments,diagnoses, and medications received, as well as the timing of these servicesor diagnoses. Previous studies have found substantial agreement between Medicaidclaims-based diagnoses of schizophrenia and clinical interviews (A.F.L., unpublisheddata, 2002) and chart reviews.17 Thus, usingMedicaid administrative claims to develop a cohort of enrollees with schizophreniahas demonstrated validity.

We used the Medicaid membership files to determine the race, sex, Medicaideligibility category, Social Security Disability status, and date of birth.We excluded enrollees who were dually eligible for Medicare and Medicaid becauseMedicaid claims would not contain complete service utilization records forthis subpopulation. Enrollees diagnosed as having schizophrenia who ever receiveda diagnosis of substance use disorder were considered to have a substanceuse disorder comorbidity. Substance use disorder diagnoses in International Classification of Diseases, Ninth Revision that wereincluded were the alcohol and drug psychoses (codes 291 and 292) and otheralcohol and drug abuse diagnoses with the exception of tobacco and antidepressantabuse (codes 303, 304, 305.0, 305.2-305.7, and 305.9).

SCHIZOPHRENIA COHORT

In an attempt to balance minimizing the false-positive and false-negativerates, the following diagnostic algorithm was used to define the study cohort.Enrollees with at least 2 schizophrenia diagnoses (codes 295.0-295.9) wereconsidered to have schizophrenia. This accounted for nearly 88.5% (N = 29 952)of the population who received any schizophrenia diagnosis in the claims data.Sensitivity analyses were conducted for the population who received a singleinpatient discharge schizophrenia diagnosis (n = 794). Of those, an additional611 did not have any diagnoses of bipolar disorder, and they were thereforeincluded in this study sample. For enrollees who received only 1 outpatientdiagnosis of schizophrenia (n = 1849), we were concerned that the diagnosismay have been formulated with less observation of a patient than that basedon a full inpatient stay, possibly resulting in more false-positive results.We also did not want to exclude the most difficult-to-engage patients withschizophrenia who may have occasionally showed for treatment. Therefore, forenrollees with a single outpatient diagnosis of schizophrenia, we conductedsensitivity analyses in which we varied the "threshold" (ie, the percentageof total mental health outpatient claims that the single schizophrenia claimrepresented) between 10% and 50%. Use of the strictest criteria resulted in541 not meeting the 50% threshold, and they were excluded from the study population.Therefore, 94.2% (n = 31 871) of the state's Medicaid enrollees who werediagnosed as having schizophrenia met our diagnostic study criteria. Enrolleesaged 18 through 64 years who met the diagnostic criteria and resided in thecomparison and carve-out regions were included.

As noted above, the people identified as having schizophrenia had tobe continuously enrolled to be included in this cohort. Continuous enrollmentwas defined as the following: per fiscal year, months not enrolled in Medicaidplus months in a Medicaid HMO must be less than 6 months.

DEPENDENT VARIABLES: QUALITY OF CARE MEASURES

Lehman et al16 conducted exhaustive reviewsof the treatment outcomes literature for schizophrenia and published evidence-basedrecommendations for pharmacologic and psychotherapeutic treatments. ThesePORT treatment recommendations were rated according to the level of evidencesupporting them. The PORT recommendations are therefore a useful tool to describethe quality of care received by enrollees diagnosed as having schizophrenia.Process measures observable in claims data were selected from the PORT recommendations,characterized as dichotomous variables and analyzed at the person–fiscalyear level. They were the receipt of (1) any antipsychoticmedication; (2) a first-generation antipsychotic medication; (3) a second-generationantipsychotic medication (not including clozapine); (4) clozapine; (5) "anti–extrapyramidalsymptom" medication, conditional on receiving a first-generation antipsychoticmedication; (6) family therapy (Current Procedural Terminology [CPT] codes 90846, 90847, 90848, and 90849);(7) individual therapy (state-specific Medicaid codes and CPT codes 90804, 90843, 90805, 90810, 90855, 90811, 90875, 90806, 90844,90807, 90808, 90842, 90809, 90812, 90813, 90814, 90815, and 90876); and (8)group therapy (state-specific Medicaid codes and CPT codes90853 and 90857). The PORT recommends that anti–extrapyramidal symptommedications should be determined on a case-by-case basis. Therefore, the absoluteproportion of persons who receive an anti–extrapyramidal symptom medicationis less important for this analysis than whether the carve-out is associatedwith changes in the prescribing of these medications.

Psychosocial rehabilitation was not recommended by the PORT becauseof an inadequate evidence base, but has face validity as being a helpful treatment.Psychosocial rehabilitation (state-specific Medicaid codes for psychosocialevaluations, basic living skills training, rehabilitation, and social rehabilitation,as well as CPT codes 97003 and 97004 for occupationaltherapy) was included in this analysis to determine whether the carve-outresponded differently to treatments with a stronger or weaker evidence baseto support them.

In addition, the PORT recommends assertive community treatment, whichincludes, but is not limited to, case management. There is no specific procedurecode for assertive community treatment, and "case management" procedure codescould reflect a heterogeneous set of services that are not specifically PORTrecommended. Therefore, we did not include case management as a quality measurein this analysis. In addition, the PORT recommends vocational rehabilitationfor patients who want to work and meet specific clinical and work historycriteria. Because those clinical characteristics are unknowable in claimsdata, this measure was not examined.

EXPLANATORY VARIABLES

Explanatory variables included age, sex, race, Medicaid eligibilitycategory, Social Security Insurance status, and presence of substance usedisorder comorbidity. We also controlled for the months enrolled in the program(independent of carve-out status) in a given fiscal year. All variables weremeasured as dummy variables except age and months enrolled. The differencein difference analysis used dummy variables for region (ie, carve-out vs comparisonregions) and time (ie, before vs after carve-out), and an interaction termbetween the two served as the difference in difference estimator. Thus, theestimator is the difference in the before and after periods in the regionthat adopted the MBHCO relative to the difference in the before and afterperiods in the control regions. The strength of this design is that it allowsus to factor out differences in utilization that are independent of the carve-outbut might otherwise be attributed to it, such as baseline utilization differencesbetween the regions.

STATISTICAL ANALYSIS

Bivariate summary statistics were computed by region and before–vs after–carve-out period; unpaired, 2-tailed t testswere used for continuous variables and Wald χ2 for categoricalvariables. The difference in differences model was estimated by means of logitregressions for each of the dependent quality measures. To account for clusteringand autocorrelation in the context of nongaussian error terms, we used thegeneralized estimating equation. Thus, the estimated standard errors are robustto autocorrelation and clustering.18

Initial analyses showed that the schizophrenia populations in the carve-outand comparison regions changed differently during the after–carve-outperiod in 2 ways: (1) the comparison regions more than doubled in their schizophreniaperson-year population, whereas the carve-out region grew only 7% relativeto the before–carve-out period, and (2) after carve-out, the comparisonregions differed in their populations based on Medicaid eligibility categories(ie, Social Security Insurance, Aid to Families With Dependent Children, orotherwise eligible). The regions also differed in their ethnic compositionin both before and after periods. Therefore, instead of comparing the beforeand after periods for all schizophrenia-diagnosed enrollees in these regions,we matched enrollees in the before period on Medicaid eligibility categoryand race, and similarly matched enrollees in the after period. Table 1 shows the results of that matching. After matching, thepopulations were similar in both matched and unmatched characteristics forthe before and after periods.

Table Graphic Jump LocationTable 1. Population Characteristics by Region Before and After Carve-Out*
QUALITY MEASURES BY REGION

Table 2 describes, aftermatching, the person-year frequency with which the schizophrenia-diagnosedenrollees received any care consistent with each quality indicator. Regardlessof region or study period, more than 85% of these enrollees received someantipsychotic medication. In contrast, very few enrollees (<1%) in anyregion received family therapy. Before carve-out, more schizophrenia-diagnosedpatients in the carve-out region received individual therapy (63.9% vs 48.3%),group therapy (32.9% vs 21.4%), or either (71.7% vs 56.5%) compared with thosein comparison regions. The opposite was true for psychosocial rehabilitation:39.7% in the carve-out region vs 48.8% in the comparison regions. In contrast,after carve-out, schizophrenia-diagnosed patients in comparison regions hadhigher frequencies of all of the PORT-recommended psychosocial treatments;they were still at less than 50% of the population (individual therapy, 16.4%in carve-out region vs 26.5% in comparison regions; group therapy, 8.4% incarve-out region vs 17.9% in comparison regions; individual or group therapy,20.3% in carve-out region vs 36.9% in comparison regions). Also, after carve-out,psychosocial rehabilitation was more likely to be received by schizophrenia-diagnosedenrollees in comparison regions than those in the carve-out (42.9% vs 16.4%,respectively).

Table Graphic Jump LocationTable 2. Quality Measures by Region, Before and After Carve-Out (Person-YearsFrequencies)*
IMPACT OF CARVE-OUT ON QUALITY

Table 3 reports the differencein differences results of each of the logit regressions. Separate models wererun for the outcome measure of all second-generation antipsychotics (ie, includingclozapine) and all second-generation antipsychotic medications excluding clozapine.The results for both models were quite similar for the 2 specifications. Theestimated coefficients for the difference in differences estimator impliesthat people enrolled in the carve-out program had the same likelihood of beingtreated with antipsychotic medications—including the more expensivesecond-generation antipsychotic medications—as those who were treatedin the comparison regions (any antipsychotic: odds ratio [OR], 1.02; 95% confidenceinterval [CI], 0.81-1.29; second-generation antipsychotic [not including clozapine]:OR, 1.05; 95% CI, 0.85-1.29; second-generation antipsychotic [including clozapine]:OR, 1.05; 95% CI, 0.86-1.28). Also, the carve-out was not associated withchanges in the likelihood of receiving anti–extrapyramidal symptom medications,conditional on receiving a first-generation antipsychotic (OR, 1.36; 95% CI,0.84-2.19). In contrast, schizophrenia-diagnosed persons in the carve-outregion were one-fourth to one-fifth as likely to receive any PORT-recommendedindividual therapy (OR, 0.27; 95% CI, 0.22-0.33), group therapy (OR, 0.19;95% CI, 0.14-0.25), or either (OR, 0.20; 95% CI, 0.16-0.24) than otherwisesimilar enrollees not enrolled in the carve-out. Medicaid recipients diagnosedas having schizophrenia were about a third as likely to receive any psychosocialrehabilitation (not a PORT recommendation) in the MBHCO region (OR, 0.31;95% CI, 0.26-0.38). Family therapy claims occurred too infrequently to beused as a dependent measure.

Table Graphic Jump LocationTable 3. Effect of Carve-Out on Receiving Quality Measures
LIMITATIONS AND STRENGTHS

This analysis relies on diagnoses based on administrative data to determineits cohort. While the gold standard for diagnosis is a structured clinicalevaluation, comparisons between Medicaid claims-based diagnoses of schizophreniawith clinical interviews and chart reviews17 havedemonstrated substantial agreement between them. Thus, claims-based schizophreniadiagnoses appear to have adequate validity for case finding. Furthermore,this schizophrenia cohort was constructed in an effort to maximize inclusionof all enrollees who were true positives (and minimize false positives) forthe diagnosis. We used a "confirmatory diagnosis" in the claims data and alsoestablished criteria so as not to exclude those who received only 1 diagnosisof schizophrenia because they were not successfully engaged in treatment.However, there is evidence that people who meet criteria for schizophreniaby clinical examination may not be diagnosed as such in the claims data. Thus,while we have likely maximized the true positives, we are less confident ofthe false negatives who did not meet these cohort criteria and therefore arenot included in the study.

These results are consistent with previous studies documenting thatfamily and group therapies are not extensively used in the treatment of schizophrenia,independent of managed care.19,20 Theyare also consistent with the PORT conformance study results21,22 thatshowed considerably higher rates for antipsychotic medications than for psychosocialtreatments. While conformance to PORT-recommended antipsychotic use is similarin both studies (87.1% vs 89.2%-92.3% in the PORT), the outpatient psychotherapiesdiffer: in fiscal years 1994 to 1996, the overall proportion receiving individualand/or group psychotherapy was 61.4% in this study compared with 45.0% inthe PORT conformance study, but after 1996 the proportion decreased to 28.6%overall. Family therapy was also more prevalent in the PORT conformance studythan in this population (<1% vs 9.6% in the PORT). However, these dataare not directly comparable. The PORT used chart review data and clinicalinterviews, whereas this study relies only on claims data. Similarly, whileapproximately 85% of all beneficiaries with schizophrenia in our analysiswere treated with an antipsychotic medication, unlike the PORT, we did notassess whether they received appropriate doses for adequate durations. Thisanalysis used a far looser indicator of quality: whether a patient received any prescription for antipsychotics in a given fiscal year.Similarly, our therapy quality measures were defined as whether these patientshad at least 1 session in a given fiscal year; we did not assess the "dose"or duration of therapy utilization. While in general it is difficult to defineexactly what is an appropriate level of utilization, it is reasonable to expectthat a high proportion of patients with schizophrenia receive at least one of these services (ie, family therapy, group or individualtherapy).

The low rate of family therapy codes in Medicaid (independent of regionor time) does not necessarily mean that family members were uninvolved intreatment or received no services. Dixon et al23 intervieweda randomly selected sample of Medicaid patients from one state and found that30% of those with ongoing family contact reported their family received informationabout their illness, treatment, or support and advice. However, only 8% respondedthat their family attended an educational or support program. Analysis ofMedicaid claims showed that 7.1% of the patients diagnosed as having schizophrenia(including those with and without ongoing family contact) had a claim forfamily therapy services. Thus, while claims data likely do not fully reflectthe receipt of any education, support, or advice, they do appear to reflectfamily therapy of an intensity that is most likely to be consistent with PORTrecommendations.

It is important to note particular limitations of psychosocial processmeasures identified through claims data. First, the PORT recommends individual,group, and family therapies that are well defined and contain specific contenttailored to an individual's or family's needs. The specific content of psychosocialtreatments is not knowable in administrative claims, nor is such informationeasily or accurately assessed through patient interview or chart review.21 As in all quality assessment of usual care, variationsin quality of a specific service are expected because of a combination ofvarying levels of clinician skill, knowledge of evidence-based care, and adequacyof resources. While possible, it is unlikely that the reduction in individualand group psychotherapies associated with the carve-out represents only cutsin therapies that were not conforming to PORT content recommendations.

Recently, Dickey et al24 also examinedthe impact of an MBHCO on the receipt of PORT-recommended quality measuresfor schizophrenia-diagnosed Medicaid enrollees in Massachusetts and foundno differences between the MBHCO and fee-for-service groups. That study isnot directly comparable to ours. In our analysis all non–dually eligible(ie, Medicaid and Medicare) recipients diagnosed as having schizophrenia wereeligible; the analysis by Dickey et al included those dually eligible andalso only those who presented for care in a crisis during the study period.Utilization in our study was determined solely by claims data and was consideredfor each person-year; in the study by Dickey et al, utilization was determinedby clinical interview and/or claims data and only for the first 6 months afterpresenting for treatment in crisis. Also, significantly, our study used adifferent design: we used a quasi-experimental design that allowed us to controlfor trends independent of the carve-out, whereas theirs was cross-sectional.Finally, Dickey et al examined a different set of contractual relationshipsbetween the MBHCO and the state. In Massachusetts, the MBHCO had limited riskof financial gain or loss. In our study, the MBHCO assumed full financialrisk for all nonpharmacologic mental health treatments. Despite these differencesin determination of population and utilization, study design, and local context,in the one quality measure directly comparable between studies (ie, any antipsychoticuse), both analyses indicated similar levels of PORT conformance.

Finally, this study found that the likelihood of receiving the nonclozapinesecond-generation antipsychotic medications increased by approximately 250%during the study period, independent of the carve-out. However, it is unclearwhether this penetration rate is appropriate. Current recommendations arethat persons who are stable with first-generation antipsychotic medicationand not experiencing significant adverse effects should not be switched tosecond-generation antipsychotics.25 Therefore,in the absence of knowing other clinical details of these patients, it isimpossible to comment on the appropriateness of this observed prescribingpattern.

POLICY IMPLICATIONS

The results of the analysis of the quasi-experiment suggest that forcontinuously enrolled non-HMO Medicaid recipients diagnosed as having schizophrenia,the MBHCO was associated with a sharp reduction in the likelihood of receivingany individual and/or group therapy and psychosocial rehabilitation. In fact,the reductions in the probability of receiving individual and/or group therapywere greater than for the probability of receiving psychosocial rehabilitationtreatments—despite the fact that these latter treatments have greaterclinical therapeutic uncertainty. The results also indicate that the carve-outdid not affect the likelihood of being treated with medications indicatedfor the treatment of schizophrenia—even the newest, most expensive antipsychoticmedications that offer therapeutic advantages over first-generation antipsychotics.Of importance, prescribing these medications had no direct economic consequencefor the carve-out vendor.

The analyses described herein lead us to a number of observations withimplications for policy. First, the data suggest that there are some clearindications of shortcomings in the quality of care for schizophrenia thatare independent of whether the mental health care was carved out. The lowrates of adoption of measures consistent with evidence-based practices suchas individual/group and family therapy are striking.

Carve-out programs respond to the incentives contained in their contracts.Our analysis shows that, in an area where strong financial incentives wereimplemented, substantial reductions in quality were observed. In contrast,when the carve-out was not financially at risk, we saw no decrement in quality.Two policy issues stem from these observations. One is that the use of high-poweredincentives to contain costs of caring for disadvantaged and vulnerable populationswill likely yield substantial savings but appears to also result in importantreductions in the quality of care for people with schizophrenia, particularlyif the contract does not include quality performance standards. Contractswith weaker financial incentives for the carve-out vendor may offer a moreattractive trade-off between cost containment and quality.

A second policy issue concerns efforts to improve quality through publicreporting. The results reported herein indicate that quality problems occurwhere there are high-powered incentives in place. In the mental health field,where carve-out arrangements virtually never include incentives to reduceprescribing of psychotropic medication, quality assessment efforts that focuson pharmacotherapies may not be measuring the therapeutic services that aremost at risk. This means that quality-of-care measurement efforts should includemeasuring treatments that are likely to be affected by incentive arrangements.

These results also raise questions regarding the argument that MBHCOsuse specialized expertise in the rationing of MH/SA care. In the wake of thesefinancial incentives, the specialty mental health carve-out arrangement wasassociated with sharp cuts in psychosocial treatments that have a known evidencebase supporting them, as well as those without such an evidence base. Theseresults point to limitations in the use of evidence-based practice recommendationsin the face of high-powered financial incentives.

Finally, while there were clear treatment quality problems in this populationthat appear to be independent of the carve-out, quality problems in usual-carepractice are not limited to one state, service system, disorder, or medicaldiscipline. The American health care system does not do well with complex,chronic conditions.2629 Addfinancial stress to the system (as is often the case, particularly in thepublic sector) and conditions can worsen. Full-risk contracts clearly reducecost but, without adequate accountability, run the risk of deteriorating qualityof care for vulnerable patients. Improving quality requires quality assessmentas well as accountability that penalizes for poor-quality care. By "payingfor quality," policymakers can realign the financial incentives to promotequality standards, not just cost containment.

Corresponding author and reprints: Alisa B. Busch, MD, MS, McLeanHospital, Proctor Bldg, 115 Mill St, Belmont, MA 02478 (e-mail: abusch@hcp.med.harvard.edu).

Submitted for publication May 19, 2003; final revision received September23, 2003; accepted November 19, 2003.

This study was supported by grants R01MH62028 and R01MH59254 from theNational Institute of Mental Health, Bethesda, Md, and the Dr Ralph and MarianFalk Medical Research Trust, Chicago, Ill (Dr Busch). The funding organizationshad no role in the collection, analysis, or interpretation of the data. Theyalso had no role in the preparation, review, or approval of the manuscript.

An earlier version of this analysis was presented as a poster at theAcademy Health Annual Research Meeting; June 25, 2002; Washington, DC.

Substance Abuse and Mental Health Services Administration (SAMHSA)., SAMHSA Managed Care Initiative State Profiles.  Washington, DC Substance Abuse and Mental Health Services Administration,US Dept of Health and Human Services1999;
Shern  DLRobinson  PStiles  PBoothroyd  RGiard  JMurrin  MRSnyder  KChen  HMassey  TBoaz  TDow  MWard  JArmstrong  MI Evaluation of Florida's Prepaid Mental Health Plan:Year 3 Report.  Tampa University of South Florida2000;
Shern  DLGiard  JRobinson  P  et al.  Evaluation of Florida's Prepaid Mental Health Plan:Year 4 Report.  Tampa Louis de la Parte Florida Mental Health Institute, Universityof South Florida July2001;
Mechanic  DMcAlpine  DD Mission unfulfilled: potholes on the road to parity. Health Aff (Millwood). 1999;1857- 21
PubMed
Frank  RGMcGuire  TG The economic functions of carve-outs in managed care. Am J Manag Care. 1998;4supplSP31- SP39
PubMed
Frank  RGMcGuire  TG Economics and mental health. Culyer  ANewhouse  JedsHandbook of HealthEconomics. Amsterdam, the Netherlands Elsevier2000;893- 954
Mechanic  D Managing behavioral health in Medicaid. N Engl J Med. 2003;3481914- 1916
PubMed
Young  ASKlap  RSherbourne  CDWells  KB The quality of care for depressive and anxiety disorders in the UnitedStates. Arch Gen Psychiatry. 2001;5855- 61
PubMed
Bloom  JRHu  T-WWallace  NCuffel  BHausman  JWSheu  MLScheffler  R Mental health costs and access under alternative capitation systemsin Colorado. Health Serv Res. 2002;37315- 340
PubMed
Mechanic  DMcAlpine  DD Utilization of specialty mental health care among persons with severemental illness: the roles of demographics, need, insurance and risk. Health Serv Res. 2000;35277- 292
PubMed
Manning  WGLiu  CFStoner  TJGray  DZLurie  NPopkin  MChristianson  JB Outcomes for Medicaid beneficiaries with schizophrenia under a prepaidmental health carve-out. J Behav Health Serv Res. 1999;26442- 450
PubMed
Callahan  JJShepard  DSBeinecke  RHLarson  MJCavanaugh  D Mental health/substance abuse treatment in managed care: the MassachusettsMedicaid experience. Health Aff (Millwood). 1995;14173- 184
PubMed
Dickey  BNormand  SLTNorton  ECAzeni  HFisher  WAltaffer  F Managing the care of schizophrenia: lessons from a 4-year MassachusettsMedicaid study. Arch Gen Psychiatry. 1996;53945- 952
PubMed
Cuffel  BJBloom  JRWallace  NHausman  JWHu  TW Two-year outcomes of fee-for-service and capitated Medicaid programsfor people with severe mental illness. Health Serv Res. 2002;37341- 359
PubMed
Ray  WADaugherty  JRMeador  KC Effect of a mental health "carve-out" program on the continuity ofantipsychotic therapy. N Engl J Med. 2003;3481885- 1894
PubMed
Lehman  AFSteinwachs  DMCo-Investigators of the PORT Project, Translating research into practice: the Schizophrenia Patient OutcomesResearch Team (PORT) treatment recommendations. Schizophr Bull. 1998;241- 10
PubMed
Lurie  NPopkin  MDysken  MMoscovice  IFinch  M Accuracy of diagnoses of schizophrenia in Medicaid claims. Hosp Community Psychiatry. 1992;4369- 71
PubMed
Zeger  SLLiang  KY Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42121- 130
PubMed
Rosenheck  RADesai  RSteinwachs  DLehman  A Benchmarking treatment of schizophrenia: a comparison of service deliveryby the national government and by state and local providers. J Nerv Ment Dis. 2000;188209- 216
PubMed
Yanos  PTCrystal  SKumar  RWalkup  JT Characteristics and service use patterns of nonelderly Medicare beneficiarieswith schizophrenia. Psychiatr Serv. 2001;521644- 1650
PubMed
Lehman  AFSteinwachs  DMCo-Investigators of the PORT Project, Patterns of usual care for schizophrenia: initial results from theSchizophrenia Patient Outcomes Research Team (PORT) Client Survey. Schizophr Bull. 1998;2411- 20discussion, 20-32.
PubMed
Lehman  AF Quality of care in mental health: the case of schizophrenia. Health Aff (Millwood). 1999;18552- 65
PubMed
Dixon  LLyles  AScott  JLehman  APostrado  LGoldman  HMcGlynn  E Services to families of adults with schizophrenia: from treatment recommendationsto dissemination. Psychiatr Serv. 1999;50233- 238
PubMed
Dickey  BNormand  SLHermann  RCEisen  SVCortes  DECleary  PDWare  N Guideline recommendations for treatment of schizophrenia: the impactof managed care. Arch Gen Psychiatry. 2003;60340- 348
PubMed
Marder  SREssock  SMMiller  ALBuchanan  RWDavis  JMKane  JMLieberman  JSchooler  NR The Mount Sinai conference on the pharmacotherapy of schizophrenia. Schizophr Bull. 2002;285- 16
PubMed
 Executive summary. Adams  KCorrigan  JMedsPriority Areas forNational Action: Transforming Health Care Quality. Washington, DC National Academy of Sciences September2003;1- 14Available at:http://books.nap.edu/execsumm_pdf/10593.pdf
Beckles  GLEngelgau  MMNarayan  KMHerman  WHAubert  REWilliamson  DF Population-based assessment of the level of care among adults withdiabetes in the U.S. Diabetes Care. 1998;211432- 1438
PubMed
Burwen  DRGalusha  DHLewis  JMBedinger  MRRadford  MJKrumholz  HMFoody  JM National and state trends in quality of care for acute myocardial infarctionbetween 1994-1995 and 1998-1999: the Medicare Health Care Quality ImprovementProgram. Arch Intern Med. 2003;1631430- 1439
PubMed
Stolar  MWEndocrine Fellows Foundation Study Group, Clinical management of the NIDDM patient: impact of the American DiabetesAssociation practice guidelines, 1985-1993. Diabetes Care. 1995;18701- 707
PubMed

Figures

Tables

Table Graphic Jump LocationTable 1. Population Characteristics by Region Before and After Carve-Out*
Table Graphic Jump LocationTable 2. Quality Measures by Region, Before and After Carve-Out (Person-YearsFrequencies)*
Table Graphic Jump LocationTable 3. Effect of Carve-Out on Receiving Quality Measures

References

Substance Abuse and Mental Health Services Administration (SAMHSA)., SAMHSA Managed Care Initiative State Profiles.  Washington, DC Substance Abuse and Mental Health Services Administration,US Dept of Health and Human Services1999;
Shern  DLRobinson  PStiles  PBoothroyd  RGiard  JMurrin  MRSnyder  KChen  HMassey  TBoaz  TDow  MWard  JArmstrong  MI Evaluation of Florida's Prepaid Mental Health Plan:Year 3 Report.  Tampa University of South Florida2000;
Shern  DLGiard  JRobinson  P  et al.  Evaluation of Florida's Prepaid Mental Health Plan:Year 4 Report.  Tampa Louis de la Parte Florida Mental Health Institute, Universityof South Florida July2001;
Mechanic  DMcAlpine  DD Mission unfulfilled: potholes on the road to parity. Health Aff (Millwood). 1999;1857- 21
PubMed
Frank  RGMcGuire  TG The economic functions of carve-outs in managed care. Am J Manag Care. 1998;4supplSP31- SP39
PubMed
Frank  RGMcGuire  TG Economics and mental health. Culyer  ANewhouse  JedsHandbook of HealthEconomics. Amsterdam, the Netherlands Elsevier2000;893- 954
Mechanic  D Managing behavioral health in Medicaid. N Engl J Med. 2003;3481914- 1916
PubMed
Young  ASKlap  RSherbourne  CDWells  KB The quality of care for depressive and anxiety disorders in the UnitedStates. Arch Gen Psychiatry. 2001;5855- 61
PubMed
Bloom  JRHu  T-WWallace  NCuffel  BHausman  JWSheu  MLScheffler  R Mental health costs and access under alternative capitation systemsin Colorado. Health Serv Res. 2002;37315- 340
PubMed
Mechanic  DMcAlpine  DD Utilization of specialty mental health care among persons with severemental illness: the roles of demographics, need, insurance and risk. Health Serv Res. 2000;35277- 292
PubMed
Manning  WGLiu  CFStoner  TJGray  DZLurie  NPopkin  MChristianson  JB Outcomes for Medicaid beneficiaries with schizophrenia under a prepaidmental health carve-out. J Behav Health Serv Res. 1999;26442- 450
PubMed
Callahan  JJShepard  DSBeinecke  RHLarson  MJCavanaugh  D Mental health/substance abuse treatment in managed care: the MassachusettsMedicaid experience. Health Aff (Millwood). 1995;14173- 184
PubMed
Dickey  BNormand  SLTNorton  ECAzeni  HFisher  WAltaffer  F Managing the care of schizophrenia: lessons from a 4-year MassachusettsMedicaid study. Arch Gen Psychiatry. 1996;53945- 952
PubMed
Cuffel  BJBloom  JRWallace  NHausman  JWHu  TW Two-year outcomes of fee-for-service and capitated Medicaid programsfor people with severe mental illness. Health Serv Res. 2002;37341- 359
PubMed
Ray  WADaugherty  JRMeador  KC Effect of a mental health "carve-out" program on the continuity ofantipsychotic therapy. N Engl J Med. 2003;3481885- 1894
PubMed
Lehman  AFSteinwachs  DMCo-Investigators of the PORT Project, Translating research into practice: the Schizophrenia Patient OutcomesResearch Team (PORT) treatment recommendations. Schizophr Bull. 1998;241- 10
PubMed
Lurie  NPopkin  MDysken  MMoscovice  IFinch  M Accuracy of diagnoses of schizophrenia in Medicaid claims. Hosp Community Psychiatry. 1992;4369- 71
PubMed
Zeger  SLLiang  KY Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42121- 130
PubMed
Rosenheck  RADesai  RSteinwachs  DLehman  A Benchmarking treatment of schizophrenia: a comparison of service deliveryby the national government and by state and local providers. J Nerv Ment Dis. 2000;188209- 216
PubMed
Yanos  PTCrystal  SKumar  RWalkup  JT Characteristics and service use patterns of nonelderly Medicare beneficiarieswith schizophrenia. Psychiatr Serv. 2001;521644- 1650
PubMed
Lehman  AFSteinwachs  DMCo-Investigators of the PORT Project, Patterns of usual care for schizophrenia: initial results from theSchizophrenia Patient Outcomes Research Team (PORT) Client Survey. Schizophr Bull. 1998;2411- 20discussion, 20-32.
PubMed
Lehman  AF Quality of care in mental health: the case of schizophrenia. Health Aff (Millwood). 1999;18552- 65
PubMed
Dixon  LLyles  AScott  JLehman  APostrado  LGoldman  HMcGlynn  E Services to families of adults with schizophrenia: from treatment recommendationsto dissemination. Psychiatr Serv. 1999;50233- 238
PubMed
Dickey  BNormand  SLHermann  RCEisen  SVCortes  DECleary  PDWare  N Guideline recommendations for treatment of schizophrenia: the impactof managed care. Arch Gen Psychiatry. 2003;60340- 348
PubMed
Marder  SREssock  SMMiller  ALBuchanan  RWDavis  JMKane  JMLieberman  JSchooler  NR The Mount Sinai conference on the pharmacotherapy of schizophrenia. Schizophr Bull. 2002;285- 16
PubMed
 Executive summary. Adams  KCorrigan  JMedsPriority Areas forNational Action: Transforming Health Care Quality. Washington, DC National Academy of Sciences September2003;1- 14Available at:http://books.nap.edu/execsumm_pdf/10593.pdf
Beckles  GLEngelgau  MMNarayan  KMHerman  WHAubert  REWilliamson  DF Population-based assessment of the level of care among adults withdiabetes in the U.S. Diabetes Care. 1998;211432- 1438
PubMed
Burwen  DRGalusha  DHLewis  JMBedinger  MRRadford  MJKrumholz  HMFoody  JM National and state trends in quality of care for acute myocardial infarctionbetween 1994-1995 and 1998-1999: the Medicare Health Care Quality ImprovementProgram. Arch Intern Med. 2003;1631430- 1439
PubMed
Stolar  MWEndocrine Fellows Foundation Study Group, Clinical management of the NIDDM patient: impact of the American DiabetesAssociation practice guidelines, 1985-1993. Diabetes Care. 1995;18701- 707
PubMed

Correspondence

CME
Meets CME requirements for:
Browse CME for all U.S. States
Accreditation Information
The American Medical Association is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The AMA designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditTM per course. Physicians should claim only the credit commensurate with the extent of their participation in the activity. Physicians who complete the CME course and score at least 80% correct on the quiz are eligible for AMA PRA Category 1 CreditTM.
Note: You must get at least of the answers correct to pass this quiz.
You have not filled in all the answers to complete this quiz
The following questions were not answered:
Sorry, you have unsuccessfully completed this CME quiz with a score of
The following questions were not answered correctly:
Commitment to Change (optional):
Indicate what change(s) you will implement in your practice, if any, based on this CME course.
Your quiz results:
The filled radio buttons indicate your responses. The preferred responses are highlighted
For CME Course: A Proposed Model for Initial Assessment and Management of Acute Heart Failure Syndromes
Indicate what changes(s) you will implement in your practice, if any, based on this CME course.
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).
Submit a Comment

Multimedia

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Articles Related By Topic
Related Topics
PubMed Articles