Background

At current (branded, single-source) prices, erythropoiesis stimulating agents (ESA’s) represent about one-quarter of care costs for Medicare dialysis patients, and 11 percent of care costs for commercially-insured patients

The 350,000 patient US market for outpatient dialysis is dominated by two players, DaVita (DVA) and Fresenius (FMS), with roughly 34 and 37 pct of patients respectively. Despite this dominance, DVA and FMS’ per-unit ESA costs are only about 8.6 percent below those of smaller, independent dialysis providers

As the overwhelmingly dominant US seller of ESA’s (for dialysis), Amgen (AMGN) historically has had no motive to offer lower pricing in exchange for market share. AMGN has in the past offered discounts as an incentive for providers to increase (per-patient) ESA volumes; however this has been largely obviated by the recent (1Q2011) change in Medicare’s method of reimbursing dialysis centers. Before 1Q2011, providers could bill separately for the ESA used by Medicare beneficiaries; since 1Q2011 Medicare has paid dialysis providers a single ‘all-in’ bundled payment intended to cover dialysis providers’ total costs – including the cost of ESA’s. This obviously has shifted the incentive from using more, to using (about 22 pct) less ESA, since the bundled Medicare payment is the same regardless of the amount of ESA used

Generic or ‘follow-on’ versions of ESA’s – specifically generic Epogen — will be among the first generic biologics approved in the US. Four versions are in US development; we expect the first truly interchangeable versions to enter in late 2015 (Novartis) to early 2016 (Hospira). Roche’s Mircera can enter as soon as July 2014, however because Mircera takes longer to begin working and is associated with more rescue transfusions than current versions of ESA’s (Epogen, Aranesp) used in dialysis, we do not expect Mircera to have a significant effect on the US ESA market – at least for dialysis. JNJ’s Procrit completed US dialysis trials a decade ago; however it’s not clear that JNJ has the freedom to sell Procrit for dialysis even after AMGN’s Epogen and Aranesp patents expire (2015)

Why Medicare’s approach to bundled payment for dialysis services could mean higher gross margins for DVA / FMS

Once generic ESA’s are available in the US, because of their dominant market share(s) we naturally expect DVA and FMS to be able to source these at significantly lower costs than their smaller peers. Under the new bundled payment methodology, this presents a dilemma for CMS[1], and an opportunity for DVA and FMS

Because DVA and FMS have a dominant share of dialysis procedures, and after generics enter should also have a large cost-of-ESA advantage versus other providers, if CMS resets all dialysis providers’ bundled payment rates to reflect average ESA costs, CMS would put nearly every non-DVA / non-FMS dialysis provider out of business. Clearly we doubt this will happen; rather we expect CMS to set the post-generic-ESA bundled reimbursement rates to dialysis providers at a level sufficient to ensure sufficient numbers (and ‘geographic spreads’) of providers. More specifically we believe that after generic ESA’s are available, CMS will adjust dialysis provider rates downward in light of lower ESA costs, but only to an extent that no more providers have negative gross margins than is presently the case[2]

Exhibit 1 shows the change in DVA / FMS gross profit dollars per Medicare dialysis patient on the assumption CMS lowers the bundled dialysis payment rate to reflect changes in average ESA costs – irrespective of the additional percentage of small dialysis providers that would have negative margins under this policy. Exhibit 2 shows the percent change in DVA / FMS gross profit dollars per Medicare patient on the far more likely assumption CMS adjusts dialysis payment rates so that the percentage of providers with negative gross margins remains constant. In each exhibit, the x-axis represents the percen t change in ESA (relative to current average costs) to DVA and FMS after generic ESA’s are available, and the y-axis represents the percent change in DVA and FMS’ gross profit dollars per Medicare patient. The three lines in each exhibit correspond to three different scenarios regarding the percent change in ESA costs to smaller (i.e. non-DVA, non-FMS) providers once generic ESA’s are available. Using Exhibit 2 as an example (assumes Medicare adjusts payment rates so that the percent of non-DVA / non-FMS dialysis providers with negative margins remains constant), if once generic ESA’s are available smaller (non-DVA / non-FMS) providers can buy ESA’s at a price 25% below the current average cost (middle green line), then DVA / FMS would have to source EPO at a cost 35% below the current average cost in order to break even[3] (see the dotted arrow in Exhibit 2). Or, holding all else constant, if DVA / FMS could source EPO for 45% less than the current average cost, they could expect average gross profit dollars per Medicare patient to rise by roughly one-third (see the solid black arrows in Exhibit 2)

What about commercial payors?

Medicare pays for roughly 85 percent of dialysis patients; commercial payors fund most of the remaining 15 percent

Where Medicare has moved to bundled payments (services and drugs all covered by a single payment), dialysis providers in most cases continue to bill commercial payors separately for drugs and services. Generalizing from payment trends for infused specialty drugs overall, we might reasonably assume that about half of commercial billings by dialysis providers charge for drugs at some discount (roughly 13 percent) to Average Wholesale Price (AWP[4]), and that around a third of commercial billings charge for drugs at a premium (about 9 percent) to Average Selling Price (ASP[5]). Still generalizing from overall specialty infusion payment patterns, we would assume the remaining one-sixth of commercial dialysis patients are billed under a bundled and/or capitated payment

CMS’ Office of the Inspector General (OIG) in January of 2009 surveyed dialysis providers’ true acquisition costs for ESA’s, so by comparing this figure with reimbursements based on AWP and ASP from that period, we can estimate that the large dialysis chains (i.e. DVA and FMS) made a roughly $3.49 / 1,000 units (42 pct) margin on ESA’s billed to commercial payors (Exhibit 3)

Looking forward, commercial payors also will have to decide how to reimburse dialysis providers as generic ESA’s are launched. AWP is likely to be retired as a pricing benchmark after it is largely replaced by Average Manufacturer Price (AMP[6]); however because AMP is not reportable for drugs not sold to retail settings (which ESA’s generally are not) it follows that in the period generic ESA’s are available, commercial payors’ choice of ESA pricing benchmarks may be limited to manufacturers’ list prices (Wholesale Acquisition Cost, or WAC[7]) and ASP

Because DVA / FMS are purchasing nearly three-quarters of the ESA’s used in dialysis, and are likely to have much lower pricing than smaller dialysis providers once generic versions of ESA are available, commercial payors choosing ASP as a reimbursement benchmark will have to bear in mind that smaller providers cannot purchase generic ESA’s at (or potentially even near) ASP. It’s the same dilemma faced by CMS, just in a slightly different form. Thus to keep smaller providers in commercial networks, ASP-based reimbursements would have to increase the reimbursement premium above ASP (currently about 9 percent) by an amount equivalent to DVA / FMS’ purchasing advantage – i.e. commercial reimbursement arguably would be roughly equal to small providers’ acquisition costs, plus 9 percent. By extension, this implies that any ESA purchasing advantage DVA / FMS enjoy relative to the smaller providers’ adds directly to DVA / FMS’ gross profits

Alternatively, commercial payors could reimburse ESA’s using WAC (roughly manufacturers’ list prices) as a benchmark. Using WAC is consistent with the currently dominant practice of using AWP, since WAC and AWP are effectively interchangeable (WAC = AWP / 1.2). However because generic ESA manufacturers are likely to sell at very large discounts to WAC, and because WAC offers no evidence of what those discounts are[8], WAC may over time be displaced by ASP as the preferred commercial pricing benchmark

We should emphasize that any transition of commercial ESA reimbursement to an ASP basis is likely to happen slowly. Because the relevance of dialysis drug costs to commercial payors is quite limited, the issue is not actively managed. All patients with end stage renal disease (ESRD) by law become Medicare eligible after a brief period, so the impact of dialysis patients on commercial payors’ medical costs is quite small. To put this in perspective – expected annual per-beneficiary (all beneficiaries, not just those on dialysis) costs for branded ESA’s among those with employer sponsored insurance is roughly $3.87, as compared to the more than $5,000 in total expected per-beneficiary annual medical costs

How do the likely post-generic reimbursement scenarios affect DVA / FMS’ earnings?

Answering this question requires three sets of assumptions: 1) how much cheaper will generic ESA’s be to smaller providers, as compared to these providers’ current costs for branded ESA; 2) how much lower are DVA / FMS’ costs for generic ESA’s likely to be as compared to the smaller providers’ costs; and 3) does this commercial reimbursement benchmark shift to ASP, and if so how quickly?

Regarding question 1: Of the four ESA’s that could in theory become more or less interchangeable with branded ESA in the US, we believe only two will in fact be established as interchangeable (Novartis & Hospira). Roche’s Mircera is unlikely to be interchangeable with ESA in dialysis because its dosing advantage is obviated in the dialysis setting[9], and because its longer onset to efficacy appears to underlay a higher incidence of rescue transfusions. And, as near as we can tell JNJ’s Procrit is kept from the dialysis market by the terms of their agreement with AMGN, even after follow-on versions of Epogen are approved in the US. Accordingly we expect at most three suppliers of ESA’s in the US dialysis market: AMGN, Novartis, and Hospira. As a very rough guess, we would venture to say that smaller providers should be able to buy generic ESA’s at 50 to 75 pct[10] of their current costs for branded EPO – and would acknowledge that the risk to our estimate lays to the downside, i.e. if anything smaller providers may be able to buy generic ESA’s more cheaply than we’ve assumed

Regarding question 2: DVA and FMS currently have a +/- 8.6 pct cost-of-ESA purchasing advantage relative to the smaller providers, in a setting where the sole supplier has little if any motive to offer discounts. In a market with multiple suppliers, we expect the two dominant buyers (combined 71 pct purchase share) to have a substantially larger purchasing advantage relative to the smaller providers[11] — i.e. 8.6 pct at the very least, and potentially much more – perhaps 25 pct or more. Bear in mind that DVA and FMS will be very aware of the fact that buying generic ESA’s below market average pricing is crucial to their gross margins, and as buyers with 34 and 37 pct purchase share respectively have the power to drive this point home with sellers

Regarding question 3: Because ESA costs to commercial payors are such an incredibly small percentage[12] (0.07 pct) of average medical costs per beneficiary, the amount of attention given to shifting the commercial ESA payment benchmark from AWP or WAC (where margins are higher) to ASP (where margins are lower) is now, and is likely to remain, limited. Accordingly we expect the shift to take place very slowly if at all

In Exhibit 4 we assume that the mix of commercial reimbursement benchmarks is effectively unchanged – 60 pct of commercially reimbursed ESA’s are paid under WAC (as a direct replacement for AWP), and the remainder under ASP. Each column corresponds to the smaller providers’ reduction in in ESA cost (relative to current costs) after generics are available, and each row to DVA / FMS’ reduction in ESA costs after generics are available. The values in each cell provide the 2016 earnings accretion (across all patients, Medicare or commercial) we would expect in each of the various scenarios. We exclude scenarios in which the smaller providers are buying ESA at the same savings as DVA / FMS, and have shaded the scenarios that we believe are relatively more likely. Exhibit 5 mirrors Exhibit 4, except in this case we assume that commercial reimbursement shifts entirely to an ASP basis

Clearly DVA / FMS should want the smaller providers to get the smallest discount possible (this keeps reimbursement higher) and to get the largest possible discounts for themselves (this minimizes costs and maximizes gross profit). We believe the Exhibit 4 (static commercial reimbursement benchmarks) scenarios are more likely than the Exhibit 5 (commercial benchmark shifts entirely to ASP) scenarios. By extension, we believe generic ESA’s can prove accretive to DVA’s 2016 EPS across a range of $0.51 to $1.14, implying a potential gain of 10 pct to 23 pct above current 2016 consensus of $4.95. Similarly, we believe generic ESA’s could be accretive to FMS’ 2016 EPS across a range of $0.17 to $0.39, implying a potential gain of 4 pct to 8 pct above current 2016 consensus of $4.75

Is this in the stocks?

Out-year expectations for dialysis providers’ earnings are to say the least complex, however we believe the market does not yet factor in the potential earnings accretion from generic ESA’s

The American Taxpayer Relief Act (ATRA) of 2012 brought in two major changes that lowered earnings expectations for dialysis providers. The first effect was the 2 pct reduction in Medicare payment rates (that took effect in 2013) as a result of the ‘sequester’). The second effect is a provision of the Act that requires CMS to recalibrate the 2014 bundled payment rate for dialysis to reflect ASP’s paid and (more importantly) average doses used in 2012. Before ATRA, the bundled payment rate reflected 2007 average doses, and assumed ESA costs were rising at the rate of the prescription producer price index (PPI) which they weren’t – ESA prices were growing more slowly. Because average doses fell (by +/- 22 pct) after bundled payments began, and because prices paid grew more slowly than assumed, the Government Accounting Office (GAO) showed in a December 7, 2012 report that the ESA component of the bundled reimbursement was effectively too high. ATRA relied on the GAO report in calling for CMS to recalibrate the ESA component of the bundle for reimbursements paid beginning in 2014. The potential EPS effects of the revision are highly significant (we estimate +/- $1.45 for DVA and +/- $0.49 for FMS); however the EPS effects of this policy shift arguably are fully priced into the stocks. Since the details of the Act became clear in late 2011, FMS’ 2013 (presumably because of the immediate sequester effect) and out-year (presumably because of the rebasing of the bundled payment) EPS consensuses have fallen substantially (Exhibit 6)[13]. And, because CMS issued to-the-penny details of the change in payment rate as a proposed rule on July 1, 2013[14], we have every reason to believe the negative effects of the Act are priced in

This leaves the question of whether the negative ATRA revisions since late 2011 have been moderated by the market’s expectations of earnings gains in ’16 and beyond as generic ESA’s become available. We believe the answer is no, for two reasons. First, the magnitude of negative revisions since the Act was anticipated (more than $2.00 for FMS) is substantially larger than our estimate of the effect of ATRA. Second, EPS growth from 2015 to 2016 should accelerate substantially if we’re right about the effects of generic ESA’s, but no such acceleration is reflected in consensus

The final possibility, which we cannot entirely rule out, is that the buy-side sees the generic ESA opportunity, but the sell-side does not. DVA / FMS (aka ‘Dialysis Providers) have a higher fPE on fy+3 (2015) consensus estimates than any other healthcare service subsector (Exhibit 7), and this would be consistent with, but by no means definitive proof of, a scenario in which buy-side expectations for out-year earnings are higher than sell-side consensus. Despite this, we believe the Dialysis Providers’ fPE’s exceed those of service subsector peers for fundamental reasons having more to do with the other subsectors’ out-year challenges: Drug Wholesalers, Drug Retailers, and PBMs all stand to be negatively affected by a shift in commercial pricing benchmarks away from AWP; Diagnostic Labs are at risk of falling Medicare payment rates (which here exceed commercial payment rates, something that occurs nowhere else, and has drawn the attention of the OIG); Hospitals face payor mix erosion (Medicaid will be a higher percentage of payments, and collections from commercially-insured beneficiaries are likely to fall as the average actuarial values of commercial plans fall); and, the commercial HMOs face significant disintermediation risks (the Medicaid HMOs are far better positioned, but in this cap weighted fPE summary have little effect on the subsector’s overall fPE)