Assessing and Addressing Antipsychotic Utilization Among Medicaid Youth: A Researcher-State Policymaker Collaboration (Text Version)

On September 16, 2009, Stephen Crystal made this presentation at the 2009 Annual Conference. Select to access the PowerPoint® presentation (466 KB) (Plugin Software Help).

Slide 1

Assessing and Addressing Antipsychotic Utilization Among Medicaid Youth: A Researcher-State Policymaker Collaboration

Stephen Crystal

Center for Education and Research on Mental Health Therapeutics Rutgers University Presentation for Session on Collaboration Between Researchers and State Policymakers: A Model for Healthcare Improvement AHRQ Annual Meeting September, 2009

Slide 2

AP Use in Medicaid Youth: Clinical and Policy Challenge

• Increased use of atypical antipsychotic (AAP) medications for a broadened range of patients and indications, often off-label, has raised a range of policy challenges for payers, patients and clinicians (Crystal et al, Health Affairs, 2009).
• Particular concerns about increased use among youth, with growing evidence of risks including metabolic effects, uncertainties about long-term effects on brain development. First-generation APs largely reserved for adults with schizophrenia and other severe psychotic disorders, but this changed following advent of the AAPs, perceived as much safer.
• In 2000s APs most expensive class of medication in MA.

Slide 3

AP Use in Medicaid Youth: Clinical and Policy Challenge

• Results from CERTs consensus meeting, expert survey, early data analyses and information on other mental health treatment challenges presented by mental health CERTs at Seattle meeting with Medicaid Medical Directors Learning Network.
• MMDs prioritized the issue of APs in youth as top priority for a collaborative project.

Slide 4

Annual Antipsychotic Use Rates by Gender, Age Medicaid FFS Youth* Ages 6-17 2001 - 2004

Annual rate of use as % of enrollees

* MAX all states except AZ, DE, DC, OR, NV, RI, NJ

Slide 5

Annual Antipsychotic Use Rates by Foster Care Medicaid FFS Youth* Ages 6-17 2001 - 2004

Annual rate of use as % of enrollees

* MAX all states except AZ, DE, DC, OR, NV, RI, NJ

Slide 6

Hierarchical Diagnostic Groups Among AP Users Medicaid FFS Youth Ages 6-17 *

* MAX all states except AZ, DE, DC, OR, NV, RI, NJ

Slide 7

Payer/Stakeholder Concerns

• Widespread use despite limited evidence base. Concerns about adequacy of initial assessment, monitoring, drug tx without close followup or psychosocial treatment.
• Especially high use in particular subgroups (e.g. foster care).
• Antipsychotic polypharmacy; concerns about dosing.
• Use among children under 6.
• Wide variation in use across clinics, geographical subareas without apparent clinical rationale.
• Unclear what are "best practices"-difficult to balance appropriate oversight and avoiding interference with clinical management. What types of cases might call for additional review and through what methods?
• These concerns led to the two CERTs efforts to work with experts and states on issues of youth AP use.

Slide 8

Collaborative CERTs Project with MA Medical Directors Learning Network: "AP Use in Kids"

• Grew out of Seattle MMDLN/CERTs meeting focusing on mental health challenges identified by MMDs.
• AHRQ supported initiative; benchmarking of AP prescribing patterns for kids, with states conducting analyses of their own data using common variable definitions and table shells. Concurrent effort to describe relevant state policies/program structures.
• Periodic conference calls to develop common definitions, data dictionary, procedures. Challenging problems of measurement as states use their own data in different ways and policies differ (e.g., age-specific guidelines; definition of polypharmacy).
• States discussed potential "best practices" in fall 2008 meeting at AHRQ.

Slide 9

Multistate Analysis Using MAX

• MAX data set up for efficient cross-state comparison with ability to pool data from multiple states. Extensive pre-processing includes translation of state codes into a common set of codes.
• Analysis by a single research team provides for consistent, replicable, documented coding; consistent inclusion criteria; pooling of relevant subgroups across states; ability to identify and follow up on anomalies; potential for multivariate analysis; multiple iterations of analyses; etc; all of which have been key to research analyses.
• However, pre-processing and other administrative steps add substantial time lag. Most recent data currently in researchers' hands are for 2004 with 2005 just beginning to be shipped.

Slide 10

Separate Data Analyses at State Level

• Separate analyses by individual states represent an alternative model.
• Advantages include: builds experience and expertise within states and increases likelihood of subsequent followup and integration of measurement into ongoing monitoring and quality improvement activities. For example, metrics that are effective at aggregate level may also be effective at provider and beneficiary level to identify and act on treatment patterns that suggest need for review.
• Allows use of most recent data, avoids time-consuming process of negotiating data use agreements with multiple states.

Slide 11

Separate Data Analyses at State Level

• Challenges: making sure that data from different data systems, programmed by different staff, are analyzed in comparable way. Challenges in understanding/documenting inclusion criteria; taking account of different program structures (e.g. FFS vs. managed care), differing mix of eligibility categories and diagnostic subgroups, etc.
• Resources required in distributed data analysis approaches to assure consistency, validity, comparability. Limited number of feasible iterations; cannot pool relevant subgroups (e.g. eligibility categories, ages, etc) across states.
• Tradeoff between simplicity of methods and required coding versus more-complex measures to improve validity (e.g. challenges with polypharmacy, adherence, etc.).
• Challenges in comparing data across states without adjustment/documentation of case-mix, eligibility differences, etc.

Slide 12

Data Dictionary

Data Dictionary developed to provide common definitions for:

• Demographic splits (age, gender, eligibility types, FFS/managed care, etc.).
• Dose (identify excessive dosing, by age group)
• Multiple drug exposure (define how to discern crossover and medication changes).
• Poly-prescribers (may be reflective of continuity of care)
• Diagnosis (what diagnoses do kids treated with APs receive).
• Maximal gap in days (are prescriptions being taken without interruption).
• Mental health eligibility descriptions (e.g, carve outs in MH services and FFS pharmacy services).
• Program characteristics (each State can be characterized by programs - central and local - to compare trends and practices).

Slide 13

Separate Data Analyses at State Level

• Despite limitations, conduct of analyses at state level has provided 16 states with a first look at their own utilization patterns and positioned the states well to extend analyses into examination of trends; identification of treatment patterns at provider and beneficiary level that suggest need for further examination; provided information to inform policy development; and laid groundwork for quality improvement initiatives and more systematic use of quality metrics.
• Perhaps most importantly, collaborative approach has enabled states to exchange information and insights about policies, potential best practices, and potential pitfalls.
• Meeting to discuss practices and utilization challenges, states identified too young, too many and too much as flags for followup action at beneficiary and aggregate level.

Slide 14

MMDLN Partnership Too young, too many, too much

• Use in ≤ 5 year olds
  • More than 11,000 users under age 6 identified.
  • Range: 0.02% - 0.67; mean approx .22%.
  • Several states examining options for additional reviews/approval procedures in this population.

Slide 15

MMDLN Partnership Too young, too many, too much

• Poly-pharmacy
  • Two levels: Use of multiple antipsychotics; use of multiple (4+) MH Drugs during year.
  • Analytic issues: Distinguishing intended or unintended 'true' poly-prescribing from tapered switches
  • MMDLN project uses simplified algorithm given practical limitations of distributed data analysis approach (overestimation of 'true' polypharmacy)
  • Work in NYS to develop and validate a better AP polypharmacy measure that can be used by other states.
• Also looked at too much (dosing).

Slide 16

Potential Applications by States Analyzing Their Own Utilization Data

• Examining AP use patterns within their state on ongoing basis can be useful tool to support policy development.
• Potential development and use of metrics for quality/appropriateness/need for case review, at aggregate, provider and beneficiary level. Ability to identify, followup on providers/beneficiaries with outlier patterns warranting review.
• Periodic utilization analyses can help Medicaid programs:
  • Determine the magnitude of regional variation in AP prescribing patterns in their State.
  • Engage their mental health community to examine the systematic and data-driven concerns surrounding AP prescription patterns.
  • Inform and support quality improvement efforts.
• Ability to evaluate changes in utilization patterns as new policies and programs are initiated.

Slide 17

Collaborative CERTs / MMDLN Project

• Potential further development of this collaborative effort
  • Further identification of best practices, evaluation of policy changes.
  • Further development of metrics for use at program and provider levels
  • Collaborative educational and other initiatives to improve evidence based use of medications
  • Analyses of use patterns in greater depth (e.g., use of validated AP polypharmacy measures; duration of use; multi-diagnosis patterns)
  • Initiatives to evaluate/improve adherence, particularly in subpopulations where consistent tx is key.
  • Atypical antipsychotics in adults

Slide 18

Comments

• Expanded and broadened use of APs in Medicaid youth is ongoing policy challenge, in the face of uncertain efficacy and safety for many subgroups of treated population. Collaboration has tremendous value for addressing this challenge. However, time- and labor-intensive; needs investment in communicating across planetary (Mars/Venus) divides; trust-building; much communication about data and other issues.
• This effort has benefited from enormous investment of effort and commitment from MMDs. Important to identify the right organizational level for collaboration (MMDs are very appropriate) and to continually keep lines of communication clear with higher authorities (e.g., NASMD briefings). Respecting political sensitivities and avoiding surprises are important.

Slide 19

Comments

• Not a short-term or "one-shot" effort. Best approached as a continuing process where initial efforts can be built on and initial kinks untangled. Greatest payoffs come as result of continuing collaboration.
• In some respects structured communication on practices and experience with policies is most-valuable result, at least in middle term.
• Development and use of metrics is also highly valuable; not necessary for each of 50 states to reinvent the wheel.
• To do "distributed data analysis" properly requires substantial resources and documentation to assure consistency and comparability. Need for funding sources for this work.

Slide 20

Comments

• Important to communicate about limitations and challenges in comparing analyses across states and to distinguish between "research-grade" analyses and those that narrow the degree of uncertainty, despite challenges.
• "When you've seen one Medicaid state, you've seen one Medicaid state." State level data best interpreted in light of local knowledge about beneficiary characteristics, program structures, sorting into eligibility categories (e.g. FFS versus managed care), coding issues, etc. (One example among many: coding for foster care status not necessarily consistent across states).

Slide 21

Value of Collaborative Researcher/Policymaker Initiatives

• Despite resource and other challenges, this collaborative approach has provided 16 states with a first look at their own utilization patterns and positioned the states well to extend analyses into examination of trends; identification of treatment patterns at provider and beneficiary level that suggest need for further examination; provided information to inform policy development; and laid groundwork for quality improvement initiatives and more systematic use of quality metrics.
• Collaborative approach has enabled states to exchange information and insights about policies, potential best practices, and potential pitfalls.
• Collaborative efforts, under the right circumstances, are of interest to Medicaid programs and offer great potential for improving care for a large, important and vulnerable population of beneficiaries.

Slide 22

Contact Information

Stephen Crystal, Ph.D.

Director, Center for Education and Research on
Mental Health Therapeutics

Rutgers University
30 College Avenue
New Brunswick, NJ 08901
voice 732-932-8579
fax 732-932-8592

scrystal@rci.rutgers.edu