Latest / Becker’s Healthcare Podcast / From Back Office to Bottom Line: Driving Revenue & Access in Specialty Pharmacy

Transcript

From Back Office to Bottom Line: Driving Revenue & Access in Specialty Pharmacy · Becker’s Healthcare Podcast

• 0:00
  Hello everyone. This is Erica Spicer Mason with
• 0:02
  Becker's Healthcare. Thank you so much for tuning
• 0:04
  into the Becker's Healthcare Podcast Series today.
• 0:08
  So today I'm thrilled to be joined again
• 0:10
  by Will Yin, the CEO of Mandolin,
• 0:13
  who will discuss driving revenue and access in
• 0:15
  specialty pharmacy.
• 0:17
  Will, thank you so much for being back
• 0:19
  with Becker's today. It's great to have you
• 0:20
  on the podcast again.
• 0:22
  Yeah. Thanks for having me, Erica.
• 0:24
  We're really happy to have you. And I
• 0:26
  know the last time that we met, we
• 0:27
  talked
• 0:28
  about specialty drug access and where pharmacy strategies
• 0:32
  are starting to intersect with AI and automation.
• 0:35
  So I'm excited to go a little bit
• 0:36
  deeper today on how you're viewing the evolving
• 0:38
  financial model for specialty therapies and health systems.
• 0:42
  So in that realm, what's changing in payer
• 0:45
  contracting,
• 0:46
  reimbursement and risk, and how does that impact
• 0:49
  pharmacy operations right now?
• 0:53
  So, of course, payer requirements are changing faster
• 0:56
  and faster, and they're they're only getting more
• 0:58
  complex.
• 0:59
  At the same time, health systems continue to
• 1:02
  deal with ever thinning margins.
• 1:05
  Specialty is a big potential revenue driver, but
• 1:07
  recent changes make that potentially challenging. So
• 1:11
  a few examples. Starting in 2026,
• 1:13
  CMS is gonna be reducing reimbursement rates for
• 1:15
  drug administration, HOPD sites, to align them more
• 1:19
  closely with rates paid for physician offices or
• 1:21
  non hospital sites. Of course, we also have
• 1:24
  the big beautiful bill,
• 1:26
  the IRA,
• 1:27
  three forty b rebate changes,
• 1:30
  and and generally,
• 1:32
  a theme of site of care requirement changes
• 1:34
  and payer steering, making it more difficult for
• 1:37
  systems to get HOPD level reimbursements.
• 1:40
  While generally this is a challenging and, potentially
• 1:44
  stressful time for health systems, there remains a
• 1:46
  significant opportunity for upside here as well. So
• 1:49
  to capitalize and,
• 1:52
  convert that potential liability into upside,
• 1:55
  health systems need three things.
• 1:58
  One, actual clarity into their own SOPs. So
• 2:02
  this seems like a small thing, but every
• 2:04
  time payers update policies, the rate of diffusion
• 2:07
  of that knowledge to a human team is
• 2:10
  heterogeneous,
• 2:11
  leading to wildly different performance and practices across
• 2:14
  teams.
• 2:15
  This was manageable in the old days, but
• 2:17
  now the cost of errors and the incentives
• 2:20
  by payers to drive errors means that health
• 2:23
  systems need clear documented
• 2:25
  practices
• 2:26
  that everyone consistently follows.
• 2:30
  Number two,
• 2:32
  consistent quality of labor.
• 2:34
  Once the appropriate SOPs and procedures
• 2:37
  are excavated,
• 2:38
  they're optimized, they're documented, and keep in mind
• 2:41
  that this is a recurring process that's redone
• 2:43
  nearly every quarter,
• 2:45
  in in many cases. IDNs need to,
• 2:48
  they need a labor force that consistently follows
• 2:51
  those practices
• 2:52
  and can be held accountable for costly missteps.
• 2:55
  Now that's, you know, obviously a challenging,
• 2:58
  ordeal, and our contention is that automated work
• 3:02
  is a huge advantage here.
• 3:04
  But whether or not it's an agent doing
• 3:06
  the work, the work being done in back
• 3:08
  offices
• 3:09
  need to start looking a lot more like
• 3:12
  an assembly line of consistent practices
• 3:15
  updated to exacting requirements.
• 3:18
  And finally
• 3:19
  data,
• 3:20
  these sophisticated systems, they can't persist without telemetry
• 3:23
  streaming out of the various systems that we're
• 3:26
  talking about. This takes a number of different
• 3:28
  shapes.
• 3:29
  So these could be QA and QC scores
• 3:32
  on workers for for for an accurate picture
• 3:35
  of where every workflow is in its process,
• 3:38
  but it also ripples upwards very quickly into
• 3:40
  getting into more strategic data system wide about
• 3:43
  reimbursements,
• 3:45
  patient out of pocket costs,
• 3:47
  and side of care optimizations.
• 3:50
  Well, thank you so much for outlining those
• 3:52
  three needs so clearly. So just to,
• 3:56
  share those back with listeners, sounds like health
• 3:58
  systems really need clarity into their SOPs,
• 4:01
  consistent quality of labor,
• 4:02
  and especially the third item data.
• 4:05
  And so I wanna pivot to talk a
• 4:06
  little bit more about how Mandolin is is
• 4:08
  approaching these needs. I understand Mandolin emphasizes
• 4:12
  reducing
• 4:13
  time to therapy and preventing revenue leakage,
• 4:16
  especially via process automation.
• 4:19
  So how are teams that you work with
• 4:21
  seeing real gains in both access and revenue
• 4:24
  cycle performance in using this
• 4:27
  tech? Great question.
• 4:29
  So let me take the time to therapy
• 4:30
  question first because it's, easier for us to
• 4:33
  see and track those gains. So on time
• 4:35
  to therapy,
• 4:37
  our customers are are seeing referral automation
• 4:39
  drop from several hours to around thirty minutes
• 4:43
  in most cases.
• 4:44
  In
• 4:45
  certain cases, especially with e scripts, we've seen
• 4:48
  an average complete intake time of around three
• 4:50
  minutes.
• 4:52
  Now,
• 4:53
  an important part of time to therapy or
• 4:55
  important contributors is, verifying benefits. So,
• 4:59
  calls,
• 5:00
  typically take longer and these, are typically done
• 5:03
  over the phone.
• 5:05
  And and, you know, the the calls take
• 5:06
  longer because an actual phone call needs to
• 5:08
  be made to a payer in most cases.
• 5:10
  Anecdotally, our customers tell us that even though
• 5:13
  a call takes one to two hours, it
• 5:15
  can take several days before the call is
• 5:17
  even made because of staffing bottlenecks.
• 5:20
  So our system automates these calls.
• 5:22
  So even though our BVs are still phone
• 5:24
  calls, most are complete within two hours of
• 5:26
  hitting our desks
• 5:28
  with rare outliers being still within a max
• 5:31
  of, say, twenty four hours or so.
• 5:33
  Now on the point of preventing revenue leakage,
• 5:36
  so I putting on my medication access hat
• 5:38
  here, we we wanna make sure that when
• 5:39
  a prescription is written, that the patient gets
• 5:42
  treated quickly,
• 5:43
  ideally within the four walls of our health
• 5:45
  system, and that we can get cleanly reimbursed
• 5:47
  for them.
• 5:48
  Sounds simple,
• 5:49
  but the reality of how to get there
• 5:51
  right now is incredibly convoluted.
• 5:54
  We first need to do a thorough benefits
• 5:56
  investigation to optimize
• 5:57
  site of care depending on what options the
• 5:59
  health system has.
• 6:01
  Traditionally, sites of care and service lines are
• 6:04
  fairly siloed.
• 6:06
  If a drug isn't covered via HOPD, it
• 6:08
  could easily flow out of the system to
• 6:11
  an independent AIC
• 6:12
  or a home infusion company,
• 6:14
  even if the health system has the infrastructure
• 6:16
  in house to process it.
• 6:19
  And running a cost or reimbursement analysis at
• 6:22
  the same time as
• 6:24
  identifying the most optimal site of care for
• 6:26
  the IDN,
• 6:27
  patient looking at a patient out of pocket
• 6:30
  cost,
• 6:31
  reimbursement to the IDN, and and ultimately access
• 6:34
  broadly is impractical,
• 6:36
  if not impossible,
• 6:38
  without the help of Agentic AI.
• 6:41
  What's incredibly powerful about AI at this moment
• 6:43
  is health system operations leaders can get visibility
• 6:46
  into which scripts are being written and margin
• 6:49
  on all sites of care across the system.
• 6:51
  And just like a command center for specialty
• 6:53
  drugs, AI can determine rules for how these
• 6:56
  scripts get routed to optimize for margin,
• 7:00
  patient financial clarity,
• 7:01
  and for keeping scripts within the system, which
• 7:04
  would otherwise be subject to leakage.
• 7:07
  Fascinating. And, Will, I know throughout our discussion
• 7:10
  already, you've noted complexity in these processes
• 7:14
  several times.
• 7:15
  And that certainly applies to
• 7:18
  areas like buy and bill and infusion operations,
• 7:21
  for hospital and health system pharmacy.
• 7:23
  So in your view, how do you think
• 7:24
  health system
• 7:26
  pharmacy leaders should evaluate whether to build, buy,
• 7:29
  or partner for their specialty drug access infrastructure?
• 7:33
  And to you, what criteria should matter most?
• 7:35
  Well,
• 7:38
  Well, I don't wanna speak for pharmacy leaders
• 7:40
  here, but I would imagine that the first
• 7:42
  step would be a really
• 7:44
  frank and honest
• 7:45
  appraisal of their ability to build out the
• 7:47
  underlying technology,
• 7:49
  process, training,
• 7:51
  and change management for a transition at this
• 7:53
  scale.
• 7:54
  And then then assuming that there are pharmacy
• 7:55
  leaders that believe that they have the technical
• 7:58
  teams and know how to to run this,
• 8:00
  then deciding whether this project is the highest
• 8:02
  and best use of their time.
• 8:04
  You very likely have seen the study out
• 8:06
  of MIT a few months back that 95%
• 8:09
  of all implementations of AI at the enterprise
• 8:13
  are failing to deliver value.
• 8:15
  I think that that study has been cited
• 8:17
  as an indictment of AI as a technology,
• 8:19
  but when you read the study, a few
• 8:20
  more nuanced
• 8:21
  themes
• 8:22
  emerge.
• 8:23
  The first is that businesses that buy AI
• 8:26
  from vendors are three times more likely to
• 8:29
  be successful
• 8:30
  with their implementation than those that choose to
• 8:32
  build in house.
• 8:34
  The second
• 8:35
  is an understanding that the technologies that are
• 8:37
  being deployed here and building out these automations
• 8:39
  is, generally speaking, not proprietary.
• 8:43
  Every AI business
• 8:44
  is ultimately buying its tools from OpenAI, Anthropic,
• 8:48
  and other large language model vendors.
• 8:51
  And, you know, I'm speaking generally here. There's
• 8:53
  there's an obvious nuance and domain expertise that's
• 8:56
  relevant, and we at Mandolin are particularly proud
• 8:59
  of in the specialty space, but it's certainly
• 9:02
  not like we invented
• 9:04
  AI and large language models.
• 9:06
  What ends up differentiating whether using a third
• 9:08
  party will be successful is not how
• 9:13
  beautiful,
• 9:15
  you know, ultimately the the the demo or
• 9:17
  the UI of the product is, but it's
• 9:19
  also in the plan that the vendor has,
• 9:22
  for handling
• 9:23
  the change management, the deployment,
• 9:25
  the understanding
• 9:27
  of payer requirements,
• 9:29
  buy and bill infusion,
• 9:31
  changing set of care requirements,
• 9:33
  and ultimately providing analytics and visibility throughout.
• 9:36
  So my own forecast,
• 9:38
  and you can say that I'm pitching Mandolin
• 9:40
  here, but the the truth is that I
• 9:42
  believe in this bet so much that I've
• 9:43
  built Mandolin's entire market approach around it, is
• 9:46
  that health systems are going to go with
• 9:48
  technology providers rather than trying to run this
• 9:51
  in house,
• 9:52
  and that the technology leaders that win the
• 9:54
  market are going to be domain obsessed
• 9:57
  with specialty drugs and have incredibly sophisticated and
• 10:00
  performant teams
• 10:02
  who excel in complex implementations at health systems
• 10:05
  specifically.
• 10:07
  Yeah. It's a powerful prediction, Will, and I
• 10:10
  appreciate
• 10:11
  you bringing us to this kind of note
• 10:12
  of forecasting here because I wanted to use
• 10:15
  our last few minutes together to talk a
• 10:17
  little bit more about,
• 10:19
  how leaders can use some of these tools
• 10:22
  to prepare for the years ahead and and
• 10:24
  what's coming. So
• 10:26
  we know that AI and automation can unlock
• 10:28
  information from manual workflows or workflows that were
• 10:31
  previously
• 10:32
  siloed.
• 10:33
  How can leaders use this data not just
• 10:36
  to operationalize
• 10:37
  tasks, but also to guide their strategy,
• 10:40
  their forecasting, and even contract negotiation?
• 10:43
  Yeah. I love this question,
• 10:47
  because what you're getting at here is a
• 10:49
  very fundamental shifting
• 10:51
  of the data and power asymmetry, frankly, back
• 10:55
  towards health systems and away from payers,
• 10:58
  that's enabled with agentic AI.
• 11:00
  I can't speak for other vendors in the
• 11:02
  space, but when Mandolin performs work on behalf
• 11:05
  of customers, we record every aspect of it
• 11:07
  in a way that can be easily summarized,
• 11:09
  charted, and tracked over time.
• 11:12
  This means that our customers can easily run
• 11:14
  analyses like
• 11:15
  reasons for denials, reimbursement rates by therapy, reimbursement
• 11:19
  rates across payers, across regions, so on and
• 11:22
  so forth.
• 11:24
  All all of this, leads to a very
• 11:26
  powerful data driven tool, that can be leveraged
• 11:29
  to negotiate better payer contracts for reimbursement.
• 11:33
  But, you know, it's also fair to recognize
• 11:35
  that it's it's to an extent running just
• 11:36
  to stay in place. The three forty b
• 11:39
  rebates that are kicking in next year
• 11:42
  are going to require all health systems to
• 11:44
  be even more sophisticated and data driven to
• 11:47
  be able to properly ensure that they're prepared
• 11:50
  for those stricter audit processes
• 11:52
  and ultimately to earn those reimbursements.
• 11:55
  Well, thank you so much.
• 11:57
  You've given us so many,
• 11:59
  great things to chew on,
• 12:01
  especially listeners as they listen to this episode
• 12:04
  and and,
• 12:05
  take this information and apply it at their
• 12:07
  own organizations. And
• 12:09
  kind of in that same vein, I wondered
• 12:11
  if you had any particular
• 12:13
  advice for health system pharmacy leaders who might
• 12:16
  be exploring
• 12:17
  automation for specialty therapies, but they haven't started
• 12:20
  yet. You know, is there one actionable step
• 12:22
  that they can take to get started this
• 12:24
  year?
• 12:27
  Well, I think the first thing to mention
• 12:29
  is that
• 12:30
  all things considered, we're still in the early
• 12:33
  innings of of AI automation. And, of course,
• 12:35
  you know, there's a lot of promise here,
• 12:37
  but there's also a lot of hype.
• 12:39
  So I'm deeply sympathetic to leaders who don't
• 12:41
  wanna jump in into nascent technologies,
• 12:45
  this early.
• 12:46
  At the same time, we have this sort
• 12:48
  of a frog in the pot situation where
• 12:50
  the water is just going to keep getting
• 12:52
  hotter.
• 12:53
  Kodiak reports,
• 12:54
  that write offs as a percentage of net
• 12:56
  patient service revenue are on track to increase
• 12:59
  something along the lines of 33% this year
• 13:01
  alone.
• 13:02
  Overall, bad debt jumped 17% in the past
• 13:05
  two years across health systems nationwide. So you
• 13:07
  can't just sit on the sidelines either.
• 13:09
  So what's the safe way to decide what
• 13:12
  to work on? Well,
• 13:14
  first off,
• 13:15
  I'm happy to chat through overall strategy with
• 13:17
  any of our listeners, by the way. So
• 13:18
  any pharmacy leaders, please hit me up on
• 13:20
  LinkedIn if you want some frank talk about
• 13:22
  this stuff.
• 13:23
  But the general advice is gonna be
• 13:26
  know what you want to accomplish,
• 13:29
  to have a frank self appraisal of your
• 13:31
  system's
• 13:32
  strengths, weaknesses,
• 13:34
  and ambitions.
• 13:36
  Three,
• 13:37
  find
• 13:38
  an automation vendor with deep domain expertise in
• 13:42
  the complexity of a specialty pharmacy space,
• 13:45
  and four, evaluate them hard.
• 13:47
  That's the way, ultimately, that you can actually
• 13:49
  separate the signal from noise
• 13:51
  in this moment.
• 13:53
  Wow. Fantastic advice, Will. And I appreciate you,
• 13:57
  opening up to listeners and and letting them
• 13:59
  know if they'd like to connect on LinkedIn
• 14:01
  to reach out. I I think it's good
• 14:02
  to keep this dialogue going. So I wanna
• 14:04
  thank you again for spending time with Becker's
• 14:06
  today and all of your insights.
• 14:08
  Thanks for having me, Erica.
• 14:10
  It's been great having you. And listeners, if
• 14:13
  you missed our first episode on pharmacy strategy
• 14:15
  meets AI automation,
• 14:17
  be sure to find that linked in the
• 14:19
  podcast description below.
• 14:21
  And a special thank you to Mandolin for
• 14:24
  sponsoring today's discussion.
• 14:25
  Hope you all have a great day and
• 14:27
  we look forward to having you join us
• 14:28
  for future episodes of the Becker's Healthcare Podcast
• 14:30
  Series.