Differences in Incentives between Stand-Alone Medicare Drug Coverage vs. Medicare Advantage Plans Kosali Simon Cornell University, y, Indiana University, y, and NBER CoAuthor: Kurt Lavetti PhD Candidate in Economics, Cornell University We are grateful for funding from NIH/NIA through grant #R03AG033211 • Well-known tradeoffs between drugs and non-drug healthcare costs – “Medicare today will pay for extended hospital stays for ulcer surgery. surgery That That’ss at a cost of about $28,000 $28 000 per patient. Yet Medicare will not pay for the drugs that eliminate the cause of most ulcers, drugs that cost about $500 $ a year....Drug coverage under Medicare will allow seniors to replace more expensive surgeries and hospitalizations with less expensive prescription medicine.” (Source: G.W.B., December 8th 2003 MMA signing, emphasis added) • Unclear whether there are incentives to internalize these spillovers within stand-alone drug coverage • =>MA & PDPs have incentives to approach f formulary l d design i diff differently tl 1| Motivation Regulation limits the extent to which they can act on differential incentives – In aggregate, each plan formulary must be actuarially equivalent to standard benefit • Provides formulary flexibility in treatment of specific drugs, and allowed to use management tools – Must cover at least two drugs g in each therapeutic p class,, and substantially all drugs in 6 key classes • Cost sharing is not regulated; drugs just have to be on formulary – Pl Plans are prohibited hibit d from f decreasing d i generosity it after ft openenrollment • Permitted to make formularies more generous after open-enrollment – Coverage is heavily subsidized, with open enrollment periods, lessening scope for selection – Part D reinsurance reinsurance, risk adjustments in drug and non non-drug drug reimbursements in Medicare =>Despite restrictions, plans may still have incentive to engage in selection and medical management • 2| Formulary Design Restrictions • Hypothesis 1: MA formularies are more generous than SA formularies for drugs whose use can lower non-drug medical costs • T To test t t effect ff t off medical di l managementt incentive on formulary design: – Identify drugs that should have large non-drug cost offsets (”List M” drugs) – Test whether MA plan formularies treat List M drugs more generously than do SA plans, relative to other drugs covered by each plan plan. – Focus on formularies outside open-enrollment to avoid confounding selection issues issues. • 3| Medical Management Hypothesis • Hypothesis 2: MA plans also differ from SA plans l in i the th incentive i ti to t discourage di enrollment (or to encourage disenrollment) of beneficiaries with ith high medical costs costs. • To test effect of selection incentive on formulary design: – Identify drugs commonly taken by patients with large non-drug medical costs (”List S” drugs) – Test whether MA plan formularies during open enrollment treat List S drugs less generously than do MA formularies outside open open-enrollment. enrollment. • 4| Selection Hypothesis During Open Enrollment Outside Open Enrollment MA Plans Selection, Spillovers No Selection, Spillovers SA Plans No Selection, No Spillovers No Selection, No Spillovers Australian PBS No Selection, Selection Spillovers No Selection, Selection Spillovers – ’Selection’ Selection • a plan has the incentive to select patients with lower non-drug medical costs – ’Spillovers’ • Medical Management; a plan has the incentive to internalize tradeoffs between drugs and other medical treatment options • 5| Identification Summary – Chandra, Gruber and McKnight, 2010 – Goldman and Philipson, 2007 – Fendrick F d i k ett al., l 2001 – (& many others) => useful for defining drug list M • 6| Evidence of Drug and Non-Drug Spillovers – Lustig, 2007 • Evidence that the existence of adverse selection causes i insurers to t distort di t t design d i off insurance i plans l in i Medicare+Choice, leading to social welfare loss • 7| Evidence of Insurer Selection • CMS Formulary Files – Has cost-sharing rules and restrictions for every plan, every drug – Doesn't have full cost of drug (prior to 2009) • Plan Finder Pricing Data – Has full cost of drug – Sample p of drugs g ((not universe)) • Australian formulary – Has medical management incentive but no selection incentive • Preliminary work uses only 2007 data • 8| Data • Similar overall generosity • Both MA and SA plans decrease generosity during open enrollment via formulary restrictions – MA plans are more likely to increase use of quantity prior limits and p authorization during open enrollment • 9| Summary Stats Descriptive Statistics: Plan Generosity 2007 Formulary Files Number of Plans Num of plan-drug pairs Quantity Limits Open Enrollment Non Open Enrollment Prior Authorization p Enrollment Open Non Open Enrollment Step Therapy p Enrollment Open Non Open Enrollment MAPDP 3 760 3,760 11,376,713 13.16% 14 26% 14.26% 11.46% 10.77% 11.59% 9.51% 2.01% 2.30% 1.55% PDP 2 100 2,100 8,460,255 12.88% 13 91% 13.91% 11.84% 11.96% 11.93% 12.00% 1.77% 2.17% 1.37% 2007 Planfinder Data • MA and SA plans have similar costsharing levels • Both MA and SA plans are more generous than the Australian formulary • 10| Summary Stats Average Initial Coverage Cost / Full Cost Num of plan-drug pairs MAPDP PDP 40.10% 67,717 40.80% 568,133 A tr li n FFormulary Australian r lr Number of Drugs Average g Initial Coverage g Cost / Full Cost Restricted Benefit Prior Authorization 3,876 50.36% 28.19% 28 15% 28.15% During Open Enrollment Outside Open Enrollment MA Plans Selection, Spillovers No Selection, Spillovers SA Plans No Selection, Selection No Spillovers No Selection, Selection No Spillovers Australian PBS p No Selection, Spillovers No Selection, Spillovers p • Generosityjd = b0 + b1ListMd + b2ListMd * MAj + b3MAj + ejd • Dependent variable: measure of plan generosity – Out-of-pocket p cost / Full cost of drug g • Compare MA to SA generosity for List M drugs relative to other drugs • Hypothesis is that b2 < 0 • 11| Methods-Medical Management Hypothesis • On average, integrated MA plans and SA plans are about equally generous for List M d drugs • There are huge differences in generosity between drug classes – Integrated plans cover ACE Inhibitors and Statins more generously, and cover Anticonvulsants, Antihypertensives, and Beta Blockers less generously • 12| Comparison of Integrated MAPDPs to Stand-Alone PDPs Dependent D d V Variable: i bl IInitial ii lC Coverage % off totall cost paid id outof-pocket by consumer Model 1 List M List M * Integrated Plan Integrated Plan N Coeff. S.E. Sig. -0.066 [0.003] *** 0.001 [0.009] -0.007 [0.004] * 124,425 Adj. R2 Model 2 Anticonvulsant ACE Inhibitor Antihypertensive Statin Beta Blocker Anticonvulsant * Integrated Plan ACE Inhibitor * Integrated Plan Antihypertensive * Integrated Plan Statin * Integrated Plan Beta Blocker * Integrated Plan I Integrated d Plan Pl N 0.005 Coeff. -0.121 0.104 -0.082 -0.049 -0.111 0.033 -0.095 0.079 -0.176 0.092 -0.007 0 007 S.E. [0.004] [0.006] [0.008] [0.008] [0.006] [0.012] [0.019] [0.028] [0.026] [0.019] [0 003] [0.003] 124,425 2 Adj. R 0.016 Findings--Medical Management--MIXED Sig. *** *** *** *** *** *** *** *** *** *** * During Open Enrollment Outside Open Enrollment MA Plans Selection, Spillovers No Selection, Spillovers SA Plans No Selection, No Spillovers No Selection, No Spillovers Australian PBS No Selection,, Spillovers p No Selection,, Spillovers p • Generosityjd = b0 + b1ListSd + b2ListSd * OEj + b3OEj + ejd • Dependent variable: 3 measures of plan generosity – % of dr drugs gs with ith 3 non non-price price form formulary lar restrictions restrictions: prior a authorization, thori ation quantity limits, step therapy • Only MA plans in regression • Hypothesis: b2 > 0 • 13| Methods - Selection Hypothesis Comparison of MAPDPs during Open Enrollment to MAPDPs outside Open Enrollment Dependent Variable: List S (Class 1) List S (Class 1) *Open Enrollment Open Enrollment N Adj. R2 Dependent Variable: List S (All) Li S (All) *Open List *O Enrollment E ll Open Enrollment N Adj. R2 • 14| Quantity Limits -0.0200 *** ((0.0003)) 0.0100 *** (0.0004) 0.0270 *** (0 0001) (0.0001) Prior Authorization 0.1095 *** ((0.0003)) 0.0083 *** (0.0004) 0.0210 *** (0 0001) (0.0001) Step Therapy -0.0080 *** ((0.0001)) -0.0017 *** (0.0002) 0.0076 *** (0 0001) (0.0001) 22,753,426 22,753,426 22,753,426 0.017 0.015 0.001 Quantity Limits -0.0055 *** (0.0002) 0 0186 *** 0.0186 (0.0003) 0.0219 *** ((0.0002)) Prior Authorization -0.0152 *** (0.0002) -0.0080 0 0080 *** (0.0003) 0.0230 *** ((0.0002)) Step Therapy 0.0014 *** (0.0001) 0 0008 *** 0.0008 (0.0001) 0.0073 *** ((0.0001)) 22,753,426 22,753,426 22,753,426 0.002 0.002 0.001 Findings - Selection • We test incentive differences in formulary design between MA and SA part D plans – Selection and medical management hypotheses • Preliminary results show large but mixed differences in medical management incentives, depending on drug class • Smaller and mixed differences in selection incenti es incentives • 15| Summary • More comprehensive measure of generosity • Trends across years • Test whether effects are isolated to p particular plans p • Include plan ownership structure to assess whether plan competition affects incentives • 16| Future work