Does Formulary Benefit Design Affect Opioid Use and Misuse for - PowerPoint PPT Presentation

Does Formulary Benefit Design Affect Opioid Use and Misuse for Disabled Medicare Beneficiaries? Erin Taylor, Rosalie Malsberger, David Powell, Janice Blanchard, Andrew Mulcahy, and Rosalie Liccardo Pacula AcademyHealth ARM, June 2017

Background • Misuse and abuse of prescription opioid analgesics remains an important public health and policy problem – Annual prescription opioid-related deaths increased more than 400% from 1999 to 2013 (Zhou et al. 2016) – Over one third of total annual economic burden ($78.5 billion) is due to health care cost and cost of treatment ($28.9 billion) (Florence et al. 2016) • Literature has offered numerous suggested drivers, although role of insurance has received limited attention – Clear evidence that insurance increases use of other Rx drugs (Duggan and Morton, 2011 and 2010) – Some recent evidence showing Part D insurance expansion played a role with prescription opioid misuse (Powell, Pacula, and Taylor, 2016)

Background • If insurance is a driver, than insurance benefit design may be effective tool in handling / reducing the problem, just as benefit design can deter excess utilization of other health care services – Placement of opioids on higher cost sharing tiers may reduce utilization, by extension likelihood of misuse – Use of utilization management (UM) tools, such as prior authorization, may reduce likelihood of use and misuse Purpose of this study: To assess whether Medicare Part D cost sharing and utilization management tools affect opioid use and misuse for disabled Medicare beneficiaries

Data Sources • Medicare data – FFS claims – Prescription Drug Event (PDE) data – Medicare Beneficiary Summary File (demographic information) – Part D benefit design data (cost sharing, UM tools, etc.) • Population of interest: disabled FFS Medicare beneficiaries living in CA or TX – Focus on disabled beneficiaries due to higher opioid use for this population relative to general population (Meara et al., 2016; Morden et al., 2014) • Timeframe: 2010 – 2012

Methods • Exploit Part D benefit design variation across plans to test effects of cost sharing and UM tools on: – Likelihood of opioid utilization (any opioids and opioid categories) – For opioid users, the likelihood of misuse • Measure misuse using previously validated measures – Trends and Risks of Opioid Use for Pain (TROUP) Score • ranges from 0-16 (low to very high risk), • combines: – # of long- and/or short-acting days supplied over 6 months – # of pharmacies dispensing opioid prescription for patient (0-5+) – # of opioid prescribers (0-5+)

Methods, continued • Calculate separate models by beneficiary Medicare/Medicaid dual eligibility status – Hypothesis: dual eligible beneficiaries not affected by cost sharing variation due to low, fixed copayments – Test effects of differential tier placement (e.g., different cost sharing) • We estimate OLS models with fixed effects and clustered standard errors at the plan level – Control for patient- and plan-level factors associated with use, including age, gender, and plan premium

Defining “Mean Cost Sharing” • Difficult to assess effects of cost sharing for an entire class of drugs when benefit design is not linear – Part D plans may offer deductible followed by tiered cost sharing • Our approach: calculate an “average expected price” – Calculate 30-day normalized cost sharing for opioids using plan data – Weight cost sharing for each plan by overall (across all plans) utilization of each opioid – If the deductible > $0: • Assume patients pay full (average) price of drug if deductible > $0 and the average price < deductible, or • Assume patients pay the deductible + plan cost sharing if drug price is greater than the deductible

Descriptive Statistics: Average Expected Price $300.00 ed OOP Price $250.00 $200.00 cted $150.00 age Expect $100.00 Average $50.00 $0.00 All Opioids Acet./Oxy. Acet./Hyd. Acet./Cod. Hydrom. Oxyc. Fentanyl Morphine Tram. Hyd. Bit. Phos. Hydrochl. Hydro. Sulf. Hydrochl. Opioid oid Drug Categori ories es All Benes (Users/Non-Users) Any Use TROUP > 0 TROUP = 0

Descriptive Statistics: Percent of Opioids with Quantity Limits 80.0% ts Applied ed 70.0% 60.0% ty Limits 50.0% h Quantity 40.0% 30.0% nt with 20.0% Percent 10.0% 0.0% All Opioids Acet./Oxy. Acet./Hyd. Acet./Cod. Hydrom. Oxyc. Fentanyl Morphine Tram. Hyd. Bit. Phos. Hydrochl. Hydro. Sulf. Hydrochl. Opioid oid Drug Categori ories es All Benes (Users/Non-Users) Any Use TROUP > 0 TROUP = 0

Descriptive Statistics (Beneficiary-Year) Always ys / Sometimes times Full Samp mple Never er Dual Dual Demogr ograph phics ics (n (n = 1, 1,176, 6,112 12) (n = 2 239,970) 0) (n (n = 9 936,142) 2) Female 48.5% 46.3% 49.0% Age* 50.6 (10.5) 54.0 (8.9) 49.8 (10.7) Race White 51.5% 61.8% 48.9% Black 16.6% 12.7% 17.6% Hispanic 26.1% 20.6% 27.5% Other/Unknown 5.7% 4.9% 6.0% State of Residence: CA 60.5% 44.5% 64.6% Risk Score* 1.1 (0.6) 0.97 (0.6) 1.2 (0.7) * Mean with standard deviation in parentheses.

Descriptive Statistics (Beneficiary-Year), cont. Always ys / Sometimes times Full Samp mple Never er Dual Dual (n (n = 1, 1,176, 6,112 12) (n = 239,970) 0) (n (n = 936,142) 2) Misuse Measures Any Opioid 47.7% 48.5% 47.5% TROUP Score (users)* 0.9 (1.8) 0.8 (1.6) 0.95 (1.8) Misuse Measure Components (Users Only) 90 Consecutive Days 6.6% 7.9% 6.3% # of Prescribers 2.2 (1.9) 1.95 (1.7) 2.2 (1.95) # of Pharmacies 1.6 (1.2) 1.5 (1.1) 1.6 (1.2) * Mean with standard deviation in parentheses.

Results: Cost Sharing Any Utilizati ization on TROU OUP P Score e (User sers) s) Cost st Sharing ng (effect ect of $10 increase) ase) Never Dual Dual Never Dual Dual Acetaminophen/Oxycodone 0.000 0.013* -0.052 0.063 Hydrochloride Acetaminophen/Hydrocodone Bitartrate 0.019** -0.004 0.077** 0.002 Acetaminophen/Codeine Phosphate -0.027* -0.021 -0.028 -0.099 Hydromorphone Hydrochloride 0.004 -0.003 0.009 0.029 Oxycodone Hydrochloride 0.000 -0.002** 0.009* -0.023** Fentanyl -0.003* -0.000 -0.010* 0.011* Morphine Sulfate -0.003 -0.009 0.005 -0.048 Tramadol Hydrochloride 0.000 -0.004 0.007 0.028 * Significant at p < 0.05 ** Significant at p < 0.01

Results: UM Tools Any Utilizati ization on TROU OUP P Score e (Us Users ers) Utilizati ization n Management gement Tools s Never Never (effec ect t of 10% incre rease) se) Dual Dual Dual Dual Step Therapy 0.004 0.000 0.045 -0.073 Quantity Limits -0.000 -0.002 0.002 0.009 Prior Authorization -0.007 -0.001 -0.078** -0.041 * Significant at p < 0.05 ** Significant at p < 0.01

Discussion • Our findings indicate that disabled Medicare beneficiaries are not responsive to higher opioid cost sharing – For all opioids, an increase of $10 in cost sharing associated with a 1.7% decrease in likelihood of utilization – statistically significant but small magnitude – Similar small effects when splitting across opioid categories • Beneficiaries tend to not be responsive to utilization management tools as well – Increase in prior authorization associated with reduction in TROUP score (increase of 10% assoc. with reduction of 0.078 points)

Limitations • Focus on only two states, which may limit generalizability due to any state-level policies in effect to control opioid use and misuse – Each state has a large disabled Medicare population • Low opioid costs may in turn limit the potential variation across Part D plan cost sharing • Beneficiaries may switch plans, and may do so to seek lower cost sharing for opioids – Conducted analysis that held beneficiaries’ plans constant for all three years to see if switching might have played a role, and found no effect

Policy Considerations • Lack of responsiveness to cost sharing and UM tools may be due to three factors: – Low cost opioid availability – Little variation across Part D plans in cost sharing – Relatively low use of UM tools to control opioid use and misuse in Part D • Policy options Part D plans might consider – Charging higher cost sharing for opioids (may not be feasible if opioid is already low-cost) – Imposing more restrictive utilization management tools This project was funded by a grant from the CDC/NIDA: 1U01CE002497-01

Funding Source This project was funded by a grant from the CDC/NIDA: 1U01CE002497-01