Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Daniel Williams (00:51):
Well, hi,
everyone. Welcome back to the
MGMA Ask an Advisor Podcast. I'mDaniel Williams.
Cristy Good (00:57):
And I'm Kristi
Good. Thanks for tuning in.
Daniel Williams (01:00):
Yeah, today we
are tackling a hot topic in the
medical practice world, payerdowncoding, especially the
increase in downcoding driven byautomation and AI. If you're
seeing more of your claims beingreduced without any review of
your medical records, you're notalone.
Cristy Good (01:19):
Absolutely, Daniel.
We've been hearing from MGMA
members across the country aboutthis. So, today, we're going to
walk through what's happening,why it's such a concern, and
what practices can do about it.We've got about eight key
questions lined up. So, let's gofor it.
Daniel Williams (01:35):
All right.
First question, Christy, what
exactly is payer downcoding andwhy are we seeing such a rise in
it right now?
Cristy Good (01:44):
So downcoding
happens when a payer lowers the
level of an E and M orevaluation and management code
on a claim. So say you have a99,214 and they downcode it to
99,213, reducing reimbursement.What's new is that many payers
are using automated claimsediting algorithms now with AI
(02:05):
powered by AI. And this is doingthis sometimes without actually
reviewing actually is doing thiswithout actually reviewing the
medical records. This goesagainst the CMS guidelines,
which require coding to be basedon either medical decision
making or total time.
So it's not just frustrating,but it's also noncompliant.
Daniel Williams (02:26):
All right.
Thanks for that one. So number
two, how are these AI basedsystems actually deciding which
codes to down code?
Cristy Good (02:36):
They're using
historical data, things like
patient demographics, commondiagnosis codes, and frequency
of high level billing. So if aclaim doesn't match what the
algorithm thinks is typical, itgets flagged. The problem, it
doesn't account for the realcomplexity of the patient
encounter. So if your providersare seeing complex patients or
spending more time coordinatingcare, the AI might still down
(03:00):
code based on averages.
Daniel Williams (03:03):
Okay. Okay. And
I'm still getting my head around
this because this was actuallynew information for me, so I'm
so glad you're explaining it toLet's just go to the next part
of this. And why is thisproblematic for providers beyond
just losing revenue?
Cristy Good (03:18):
It's a huge
administrative burden. So your
staff not only is trying tocatch those that are being
downcoded, but they also gatherthe supporting documentation and
they have to submit appeals. Sothey're already doing the front
end work quite often with usingtheir coders to code and
document and making sureeverything's in the right order
(03:39):
and the right documentation, andthey send it off. But if it
comes back, it they may have tosubmit appeals, which is such an
administrative burden on thepractice. It also undermines the
trust in the process.
CMS and the AMA have both madeit clear that claims should not
be adjusted without review ofmedical records. So this kind of
(03:59):
downcoding can lead to burnout,delay in payment, and even
inaccurate performance profilesfor providers.
Daniel Williams (04:06):
Okay. That's
blowing my mind because we keep
thinking AI is just makingthings easier, but in this case,
it's actually creating moreadministrative burden?
Cristy Good (04:17):
Because of what the
practice has to do to all to
appeal, even if the practice isusing some AI for help with
their coding to make sure itgoes over correctly to the
billing. They're still getting alot of denials, and then all
those appeals are just anotherlayer, and it's taken an
employee's time to go through,gather again documentation, fill
(04:40):
out the paperwork, and thenresubmit those claims.
Daniel Williams (04:44):
Okay. So.
Cristy Good (04:45):
Making it easier on
the payer.
Daniel Williams (04:47):
Yeah. Yeah.
Cristy Good (04:49):
But not the
provider.
Daniel Williams (04:51):
And that's not
what we want here. We want to
help our practices out there.So, if you are a practice
suffering from this or you wantto know if it's an issue for
you, what are the signs apractice should be looking for
to know if they're being downcoded?
Cristy Good (05:06):
I think probably
many are checking their
remittance advance. Phrases likeservice level adjusted based on
payer policy or billed serviceinconsistent with diagnosis are
common signs. Also look forpatterns like high level E and M
codes routinely paid as lowerlevels, especially by the same
payer. If one or two providersseem to be impacted more than
(05:28):
others, that's worth tracking aswell.
Daniel Williams (05:31):
Okay. Next up,
what steps then should a
practice take once they identifydowncoding?
Cristy Good (05:38):
They should check
everything. Which payer, which
codes, which providers, how manywere appealed, and how long it
took, and how many wereoverturned. Appeal every
unjustified downcode withsupporting documentation, and
send that data both to the AMAand even to our MGMA government
(05:59):
affairs team here, which wouldbe gov, G0VAFF,@MGMA.com,
because they're gonna be lookingat how many practices are being
impacted by this downcoding, andthose examples will help us
advocate for better payeraccountability.
Daniel Williams (06:19):
Okay. Do we
know right now at all how
widespread this is? Is it acrossall practices?
Cristy Good (06:25):
Well, I'm getting,
it's a number of Blue Cross Blue
Shields. And I am getting, weknow that UnitedHealthcare had a
big thing about it, about usingAI to look at their claims. I
know that our government affairssaid they are working with a
couple practices already, and wejust seem to be getting more and
more questions coming through tous saying, I'm doing all that I
(06:48):
should be doing. I need help.They're still denying.
And now it's causing me morework on our end to reappill
those.
Daniel Williams (06:56):
Okay. Let's
talk documentation then. What
can practices do to reduce theirrisk of down coding in the first
place?
Cristy Good (07:05):
So we know that
good documentation is your best
defense. So be specific, clearlyoutline your medical decision
making process, document totaltime if time based coding is
used, and make sure to includeany chronic conditions or
comorbidities that impact thevisit. Avoid copy paste template
(07:26):
language. That is a big thingthat I think causes some of this
to happen with and gives redflags to the AI system. If
you're not already doing annualexternal audits, that is
something I'd recommend.
And, of course, your internalaudits. If you're having certain
providers that seem to be moreaffected than others, definitely
(07:46):
get those audited because therecould be some documentation. You
might be thinking you're doingit right. The providers might be
thinking they're doing it right,and there might be something
that's being missed.
Daniel Williams (07:58):
Okay. Now you
raised something here that,
again, it just, it caught myattention. You said avoid copy
paste template language. Itraises red flags there. Tell me
what that is.
I can imagine what it is, butI'm not really sure in a
practice what's happening so wecan certainly try to provide
(08:19):
insight for our listeners sothey don't do this.
Cristy Good (08:22):
Yeah, it's in the
EHR. When you're just copying
your note from your previousnote, or you're using the same
templates over and over as yourdiagnosis, you, because you can
build templates in your EHR, butif you're constantly using
certain templates for all yourpatients with diagnosis and
reasoning, that it does make itlook like possibly you haven't
(08:45):
looked into that patient asdeeply as maybe you should, or
you're missing something, it'sjust another little red flag. So
even when I did some EHRimplementations, we built those
templates, but we also remindedthem not to just copy and paste
previous notes and not just touse specific templates for
(09:06):
everything.
Daniel Williams (09:07):
Okay. All
right. Thanks for sharing that
information. Let's talk aboutwhat role AI can play on the
provider side then. Canpractices use it to their
advantage?
Because you and I have talkedmany times about AI as a tool,
something we can work with tohelp us do things more
efficiently, etcetera. But I'mhearing bad news for AI right
(09:30):
here. So how do we use it to ouradvantage here?
Cristy Good (09:34):
No, absolutely. And
that's what I've even told our
members, that you combat theirAI with your AI. You just make
sure that if you're using AIdriven coding assistance or
audit tools to assess yourclaims before you submit it,
just also take a look and makesure that it's doing what it
(09:54):
should be doing. These tools canflag missing documentation or
inconsistent coding and thenhelp ensure that your claim
meets those payer expectationsbefore they get rejected or down
coded. It is a great way tolevel the playing field when we
know payers are using AI driventools too.
But just like with any AI, it ismachine learning, and you should
(10:17):
always make sure you're checkingwhat it's giving you, the
answers it's giving you, whatit's doing. And even if it's
sending clean claims over, andyou're getting denials, it's
still worth looking into.
Daniel Williams (10:31):
Okay, that
makes sense. That makes total
sense. So what can practices dofrom an advocacy or policy
standpoint to push back? I knowyou mentioned earlier about AMA.
You also talked about MGMA'sgovernment affairs teams.
I know I was on an email chainyou shared with me where you
were actually chatting with ourgovernment affairs team. So
(10:51):
let's talk about that from thatadvocacy or policy standpoint.
Cristy Good (10:56):
Sure. First, it's
great to hear from our members
out there that are experiencingthis. I suggest posting your
experience on the MGMAcommunity. If you're not
involved, it's a great way toget advice from your colleagues,
chat with your colleagues, sharethoughts with your colleagues.
And it just helps us see ifthere's some trends across
(11:17):
practices, because if morepeople are posting, then we'll
be knowing that it's not oneperson affected, it's more.
Second, report patterns to theAMA. I know the AMA has been
looking at this, so I think theyhave a downcoding survey, and I
think we we'll have a link tothe article from the AMA on this
at the end of our podcast. So inthere is a survey link that you
(11:41):
can go to. And then third, ifyou believe your payer is
violating contract terms orstate regulations, you can file
a complaint with your stateinsurance department and then
send us your data. We work withMGMA's, our government affairs
team, and they're really pushingfor transparency and fairness in
payer practices.
So the more they know, the morethey can advocate for our
(12:02):
members on behalf.
Daniel Williams (12:05):
Okay, that
sounds great. Because we've
covered a lot of information,Christy, help us just, I guess
the best thing is to, what arethree big takeaways you'd want
our listeners to know from thisconversation today?
Cristy Good (12:19):
Sure. The first
one, be vigilant. Track
everything and know your data.That is really key. You should
know what's coming in, goingout, doesn't make sense, and
track it.
Be proactive, strengthen yourdocumentation, and appeal when
needed. If you have questions ondocumentation, you can reach out
to us here at askadvisormgma dotcom or any of the other coding
(12:43):
resources out there. And then beconnected, use MGMA, use our
community, and the AMA, and yourstate channels.
Daniel Williams (12:51):
Perfect. And
everybody, as Christy said,
we're going to drop some directlinks there in our episode show
notes. And if you are lookingfor support with coding audits,
compliance tools, or payerstrategy, reach out to us at
advisormgma dot com. You canalso contact consultingmgma dot
com if you're ready to bring inadditional expert help.
Cristy Good (13:14):
Thanks for spending
time with us today, and we hope
this gives you the tools andconfidence you need to address
your repair downcoding in yourown practice.
Daniel Williams (13:22):
Until next
time, I'm Danielle Williams.
Cristy Good (13:24):
And I'm Christy
Good.