Primary Care Physician Response to a Mental Health Carve-Out: An Economic Analysis
Ashley Aull Dunham Jennifer L. Troyer William P. Brandon UNC-Charlotte
Carve-Outs
specific services from prepaid health plans “Specialists” manage benefits Distinct budgets, provider networks and incentive arrangements Effect on Primary Care Physicians?
Exclude
Project Background
Mecklenburg County (Experimental)
n=3497
Mandatory Medicaid HMO enrollment in mid-1990s Mental Health, Dental Care and Prescription Drug Carve-Out – absolved HMO of both responsibility and risk
New Hanover County (Control)
n=969
Traditional FFS
Incentive Structures
Mental Health FFS and Primary Care FFS (Control)
“Traditional” incentive to increase business volume Primary care will refer mental health only when timeconsuming or problem cases that prevent wealth maximization
Primary Care Capitation with Mental Health Carve-Out (Experimental)
Clear incentive to move mental health out of primary care Does this conflict with the patient’s best interest?
Data Available
Encounter forms not reliable Claims data for antidepressant prescriptions by all providers and visits provided by mental health professionals coded for depression
Research considers the effect of a
mandatory Medicaid mental health carve-out (that precludes reimbursement for mental health in primary care) on depression treatment for a sample of Medicaid recipients.
Methods – DID Models
Yit = β0 + β1Countyi + β2Phaseint + β3Postt + β4(Countyi)(Phaseint) + β5(Countyi)(Postt) + εi
Claims/month submitted by mental health providers coded for depression Antidepressant prescription claims/month Antidepressant prescription claims/month submitted by mental health providers Antidepressant prescription claims/month submitted by non-mental health providers
Methods – Logit Models
Pr(Yit=1) = β0 + β1Countyi + β2Carveoutit + β3(Countyi)(Carveoutit) + β4Racei + β5Genderi + β6Agegrpi + β7Categoryi + β8Timeoni +εi
Probability that a mental health provider prescribed antidepressants (as opposed to all other providers) Probability of antidepressant claims in a sample of all drug claims
Effects of the Mental Health Carve-Out (DID Models)
Mental Health Visits County Phasein Post (County)(Post) *p < .05 **p<.10 -1.96645* .74444 -1.20064** 2.31707* Total Antidepressant Claims -1.43068** 2.92664* 6.41282* -2.14449** -3.24974* Claims Prescribed by Mental Health Providers 0.86485 3.79998* 5.35710* -3.75920* -3.69659* Claims Prescribed by Non-Mental Health Providers -2.22826* -.87334 1.05572** 1.54743** .37957
(County)(Phasein) -0.15785
Effects of a Mental Health Carve-Out (Logit Models)
Pseudo R2=.2121 n=861 Antidepressant Prescribed by a Mental Health Provider .8583447 .6562034 -1.112025 1.756301* -1.238316** 2.634148* -1.130781 .0251983* Pseudo R2=.1082 n=41,477 Drug Claim Being an Antidepressant .5297924** .4082136 -.1327179 1.333435* -.2423945 1.687171* -.792415 -.0048483
County Carve-out (County)(Carve-out) Race Gender Agegrp Category Timeon *p < .05 **p<.10
Results – DID Models
Significant increase in mental health claims for depression (supports theory of wealth maximization)
Increase in referrals may have only been for severe depression – in the patient’s best interest
Decrease in antidepressant claims from mental health providers and no change in antidepressant claims from non-mental health providers
Primary care providers continued to treat for depression by prescribing (free good) – allows wealth maximization while continuing to serve the patient’s best interest
Results – Logit Models
No significant change in prescriptions for antidepressants and no change in probability that prescription came from a mental health provider
Decreased likelihood that mental health providers used antidepressants
No additional barriers to getting antidepressants relative to all other drugs by eliminating primary care reimbursement for mental health
Conclusions
Physician’s utility function defined by many factors, including wealth maximization and their obligation to serve as a perfect agent Data suggests their obligation to serve as advocate was more powerful than their need to maximize wealth
Removed from reimbursement arrangements and small portion of their patient population Less sensitive to reimbursement changes
Implementation of capitation with the Medicaid population did not cause uniform change in behavior