ICD-10 and Clinical Documentation

ICD-10 and Clinical Documentation


Hi, I’m Dr. Joe Nichols
from Seattle, Washington. Welcome to this program on “ICD-10 and Clinical
Documentation”. The goals of this program
are to: describe the role of documentation and coding
in medical care; additionally, we want
to characterize the type of information that should be
documented with ICD-10, which is actually
quite consistent with the same documentation
that is needed for good,
quality medical care. We’re going to try
to address some questions around clinical
documentation in the course
of this presentation: Why are clinical documentation
and good coding important? How can we make
ICD-10 transition easier to both the clinician
and the coder? What type of documentation
is needed for different specialties,
clinical areas, or patient conditions? What should practices
focus on, in terms of ICD-10,
in this transition? How can the provider
clarify what needs to be documented
and identify that? And why is coding
for specificity to define the key parameters of the patient
health condition important? How can the provider make sure
that documentation quality and coding quality
is achieved and maintained? So why is it important?
Well, for one thing, it supports proper payment
and reduces denials. It ensures accurate measures
of quality and efficiency; it ensures accountability
and transparency; it captures the level
of risk and severity; it provides better
business intelligence; it supports clinical
research, particularly population-based
clinical research; and it enhances communication
with hospitals, other providers. The most important reason
for good documentation is it’s just
good patient care. If we look historically
at how we’ve used these codes, one of the things that stands
out is that we’ve only used a very small percentage
of the codes, historically. This was based off of a study
that I did a couple of years ago that looked at 3 years of data,
about a million lives, for a payer that covered
all lines of business and, in this study, 72.5%
of all the charges were coded with 5%
of the codes, so we’ve traditionally only used
a very small percentage of the codes for a very large
percentage of the business. Now, this is not necessarily
a good thing, because, as we look at some
of those codes, we find that many of them
are extremely vague and it’s very difficult to tell
what actually was going on with the patient
because they tend to be highly unspecified
or symptom-type codes. As a matter fact,
in looking at the study, 54% of all the codes
over those 3 years were either unspecified, other,
or pure symptom-type codes, so, moving forward, in ICD-10, we
hope to get a lot better data. We have to change,
historically, what we’ve done, in terms of
our coding patterns, and all this goes back
to basic clinical documentation. We can’t code
what isn’t documented. If we look at the
transition from ICD-9, we also see that there
are substantial differences in different
clinical areas. The area of musculoskeletal care
is highly impacted by ICD-10: we have a lot more codes, and
there’s a good reason for that. We’ve gone from 747
fractures in 9 to almost 17,000 codes
for fractures in ICD-10. Pregnancy-related conditions
have doubled, in terms of the number of codes,
but, if you look at pregnancy-related
conditions in ICD-10, for almost every code,
there’s a separate code for 1st trimester, 2nd
trimester, and 3rd trimester. So if you look
at the number of codes, we actually should have
3 times as many codes. We actually only have
twice as many codes, which suggests that,
if we factor out the conditions of 1st trimester, 2nd trimester,
and 3rd trimester, we actually have
less codes, overall. In looking at things
like diabetes, we have more codes
because we’ve combined many of the different complications
of diabetes with the code, so we now have
a lot of codes with all of the
complications attached. In some areas,
the codes have gone down. For example,
mood-related disorders or affective disorders, the number of codes
has decreased. Hypertensive disease:
half the number of codes. End-stage renal disease:
half the number of codes. So the impact, in terms
of just the number of codes, is highly variable, depending on
the specific type of condition, and every condition
is somewhat different, as it has been in ICD-9. Just to look at why
we have so many codes, and particularly looking
at fracture codes, because that seems to be
the greatest number of codes; almost 25%
of all the ICD-10 codes are fracture-related codes. And, if we look
in the tabular index, we’ll see a section called
fracture of the femur, fracture of one bone,
for which there’s 2,466 codes. Well, that isn’t
because there’s 2,466 ways to break the femur;
it’s really because we have a lot of
different characteristics of all of these
different fractures where we describe:
is this the initial encounter, is it the subsequent encounter,
is it normal healing, delayed healing,
nonunion or malunion; is it an open fracture and,
if so, what type? And so we have a lot
of characteristics about each fracture and,
as we add those characteristics, we end up
with a lot more codes because we end up
multiplying them. If you think about CPT
and having a modifier, if we had a different code
for right and left in CPT, we’d have twice
as many codes for those things that
represent right and left, but, instead,
we have a modifier, so it’s the structure
and the nature of the codes that cause us to have
so many codes. The actual specific
medical concepts really aren’t
substantially different. We do have a lot of ability to identify key parameters
of severity and risk in ICD-10 that we couldn’t quite
identify previously in ICD-9: more specific things about
comorbidities, manifestations, causes, complications,
detailed anatomical location, and all of these things
have a huge impact, in terms of the risk
and severity and complexity
of the case. For example,
if we have a patient who has a fracture
of the right hip, if they have
a subcapital fracture, the mortality and morbidity
is substantially reduced because you can do
the operation quite quickly and you can get the patient
up the next day. Those are the two factors
that have the most impact, in terms of mortality
and morbidity, for hip fractures. For an intertrochanteric
fracture, it’s a bit more complex
and it’s a bit slower, in terms of getting
patients up; and, for a subtrochanteric
fracture, we now have a very unstable-type fracture,
where the operation is quite lengthy
and it takes a long time and a great deal of care,
in terms of getting patients up. Well, those fractures are
3 inches from each other, but the morbidity and mortality
from those fractures is dramatically different, so very small differences in
anatomy or other characteristics make a huge difference,
in terms of the risk and complexity and severity
of these conditions. Now, there’s nothing new
about clinical documentation. I pulled up a record from
the Internet, back from 1889, and this was the evaluation
of a patient who got shot in the thigh from
a musket during the Civil War and was complaining that they
had rheumatism because of this and wanted an increase
in the disability, something like
a dollar a month, which, at the time was,
of course, outlandish. Well, the physician
who performed the evaluation was very specific;
he defined the problem: gunshot wound, right thigh;
and rheumatism. All of the vital signs were
recorded with great detail. And, within the record,
he describes specifically the small superficial scars
in the midthigh and describes exactly
the date that it occurred and he also describes other
key comorbid conditions, like the fact that the patient
had an amputation of the middle finger through the
proximal interphalangeal joint, so a great deal of detail. Obviously, not
an electronic health record, but we’ve always had
documentation and, to some degree,
as we’ve looked forward, it’s been better
than we have today. So if we look at some
of the documentation we have today,
it’s pretty bad. Now, hopefully, most clinicians
are not documenting like this, but it’s still happening today
and we’re still seeing abbreviations
that no one understands, and probably including
the physician who wrote them, and, really, this is
not only bad documentation; ultimately, we can’t take
proper care of the patient, if we don’t have good
documentation of the key parameters
of that patient that are important. Where does it all start?
Well, it starts where we were taught, in medical
school and residency, where we were taught to do
an external record review, to get a history
from the patient, to perform
a physical examination, to look at internal records,
to look at studies, and to pull all
of that information together in a very specific way
and make sure we understood all the key parameters
of the patient condition, so that we could come up
with the appropriate treatment. What are those types
of things we did? It’s the same things we were
taught in medical school. What type of condition is it?
How do we categorize it? Is it type 1 diabetes,
type 2 diabetes? What was the onset? Was it
a month ago, was it 4 weeks ago? What was the etiology? Was it caused
by an infectious agent? Was there some type
of internal failure? Is it a congenital problem? What was
the anatomic location? Is it one of those
subtrochanteric fractures? Is it an intertrochanteric
fracture? Is it an injury of the
upper pole of the kidney? Is it a problem with the left
upper lobe of the lung? Is it the right side,
the left side? Is it mild,
moderate, or severe? What are the environmental
factors around that? All of these things, again,
are key clinical concepts that are important, in terms
of taking care of the patient, and all of them are
the key parameters that we were taught,
in medical school and residency, that we should capture,
where relevant, for that particular
patient’s condition. Time parameters:
is this intermittent, is it paroxysmal,
is it recurring, is it acute or chronic?
Are there comorbid conditions? If that patient has diabetes,
did they have retinopathy, did they have
a neuropathic joint? If a patient has a spine
injury or a head injury, do they have loss of
consciousness or paralysis? Healing levels: how well is it doing? Are we having normal healing?
Is it delayed? Is there a nonunion?
The fracture simply didn’t heal at all;
or did it heal, pointed in the wrong direction,
and result in a malunion? All pretty important stuff. Findings, fevers, wheezing,
external causes. What was the cause of this? Was that patient involved
in a motor vehicle accident? How does that factor into how we
handle that patient’s condition? How does it factor
into how that patient’s claim and payment
is going to be handled? There’s a lot of important
factors around that. And what type
of encounter was it: was it an initial encounter,
subsequent encounter? So we see that there’s a lot
of key different parameters. So let’s take one condition,
otitis media, a very common condition,
seen frequently in the pediatrician
and the family practice office. Now, we could look at that
patient who comes in, and the mother says the patient
just has an earache, and we could come up
with a diagnosis that simply says
what the mother said: otalgia, which basically
is just earache. Now, we really should be able
to do a bit better than that. So if the patient comes in,
we want to be able to capture, for that patient
who has otitis media, what type of otitis it is. Is it just
a run-of-the-mill type; is it a specific type
of otitis media? What was the onset?
Is this acute? Is it a chronic condition? Did the patient have measles
or influenza or allergies that may have triggered
or caused this otitis? Which side was involved:
was it the right side, left side,
was it bilateral? Were there other factors? Was there a smoking history
in the family or the home and the patient comes in
recurrently with otitis media and is exposed to environmental
tobacco on a regular basis? Probably pretty important
stuff to know. Is this recurring? Is it acute, is it chronic? Did the patient have a rupture
of the membrane, or not? I can’t imagine a patient
coming in with severe otitis and complaining of earache,
where the physician would not at least look in the ear to see
whether the membrane was intact; that’s sort of one of the basic
things that we were taught. Or is there a drainage
that’s serous or fluidlike or is it suppurative
or a purulent type of drainage? All of these things
are key factors and, if we look at ICD-10,
what we see is that a lot of these types of concepts
are captured in the code. We can now capture whether
there is a specific cause, whether it’s acute
or subacute, whether there was
rupture of the membrane, whether there is drainage
of a serous type or is a purulent type. Much better to capture
this type of information to characterize the nature
of the condition, than simply say
the patient had an earache. Now, clearly, there’s
a code for earache, but it should probably not be
the standard code we use and it should not be based
on the fact that we didn’t document anything other than
the patient had an earache. As we move into ICD-10, there’s still a lot of interest,
in terms of coding specificity. We’ve looked,
historically, at the data and the data shows that
we really have not captured a lot of information
about the patient that really
should’ve been captured and there’s a lot
of pressure, looking at “maybe we should just not accept
these unspecified codes.” Well, there’s a real challenge
in that because there’s times when unspecified codes are
exactly the right code to use. So how do we describe
poorly specified coding? So I’m going to propose
a definition, that: “Coding that does not fully
define the important parameters of the patient condition
that could otherwise be defined given the information available
to the observer (the clinician) and the coder is probably
less than optimal coding.” Coding specificity,
as we mentioned, may be very unspecified
for a very good reason. Sometimes we’ll see
the patient early in the course
of the treatment. They present in; they said,
“Doc, I have abdominal pain. I don’t know why it’s there”
and you examine the patient; there’s nothing specific
that you can identify. The patient cannot relate
what caused this or when it came on
and is very, very vague. And we see this all the time. The proper code to use might be
“abdominal pain, unspecified”. We have to ask, though,
if this is the 10th visit for the patient
and the diagnosis is still abdominal pain, unspecified,
we have to wonder whether we’re adding sufficient
value at that point in time. So coding specificity
is something that may start out very nonspecific, because that’s
all the information we have, but, assuming we evaluate that
patient and examine that patient and come up with
a more definitive diagnosis, ultimately, we should have
that more definitive diagnosis supported by
the documentation necessary to get to that
better diagnosis. We know,
moving into ICD-10, that there are a lot of codes
that are unspecified. We can essentially be
just as vague, in ICD-10, as we were in ICD-9. There are a lot of codes there
that are not terribly specific and a lot of codes that we
should rarely, if ever, use. For example, in ICD-10,
we frequently will have right side or left side and we also have codes
that say “unspecified side”. Now, we should rarely
use that; we should almost always know
our right from our left, particularly if we’re
treating our patient. How can we possibly take care
of that patient, if we don’t? We ought to know what
anatomical location it is. If we say “injury, limb”,
it’s pretty vague. Even the patient knows where
that was injured, in the limb, or what type of injury it was,
so that type of coding, at that level, should
probably never be considered, because there really isn’t
a reason for it. Or, if you’re an OB physician
treating a patient, you probably know
what trimester it is. It would be very, very rare,
if at all, that you would treat
a patient and say “trimester unspecified”,
or should. If you’re treating
a patient with diabetes, you should probably know
what type of diabetes it is. If the patient has specific
known complications or comorbidities, that
should clearly be documented and, if relevant,
should be coded. We also should be able
to say is it severe, is it acute,
is it chronic. You probably
should not say things like “respiratory failure,
unspecified”, because we never had
an unspecified code in ICD-9 and we do in ICD-10,
and because how can we treat the patient
with respiratory failure, if we don’t know whether
it’s acute or chronic? So when we start looking
at some of these codes, there are times
when “unspecified” makes a lot of sense;
as a matter of fact, we shouldn’t be
making up stuff. If we don’t know
what’s going on, we should use
the more unspecified code, but there are many times when we
use an unspecified or vague code that really doesn’t make
any sense and should rarely,
if ever, be used. Anytime we have care
that demands a more specific level
of detail, in order to care
for the patient, we should have the
documentation to support it. If we’re
at a specialty level, we should be able
to provide more detail. The primary care physician
may not know which Salter-Harris
classification that growth plate
injury was, but the orthopedist
certainly should. So we have different levels
of specificity for good reasons: Because it may be early
in the course of treatment; it may be that a physician
or some other provider who is at a different level,
who really doesn’t know all of the details,
at any point in time, or may not have the training
to recognize more detail. And then, we have coding
at a more specific detail level, which is important
for patient care, but may be coded by
a different type of provider, who actually has
more specialty training. There’s a wide range of
different levels of specificity; we should always try
and use the best specificity that’s important
for the patient care. Ultimately, what we want
to be able to get is good patient data
because good patient data is going to be the foundation
for improving healthcare: for understanding what we’re
doing with patient care, for understanding
different costs and different complexities
and different severity and different complications
and different outcomes and why some outcomes
are different than others, because the patient’s condition
is more or less severe or a different type
of condition. All of these things
are critically important as we’re moving into an era
of value-based purchasing, accountable care,
of risk adjustment, of taking on
the risk of population. Understanding the level
of the key parameters of the patient condition, their
complexity, their severity: all of those things become
increasingly more important from a financial,
from an analytic, and from a patient-care
standpoint. From my perspective, good patient care is all
about good data; good data is all about
good patient care and, if we want
good patient data, then we have to have
3 basic things. We have to have:
observations of all the facts relevant to
the patient condition. We have to have
documentation of all those
key medical concepts. As we were taught
in medical school and has been brought to
our attention many times since: “if it’s not documented,
it didn’t happen.” As a matter of fact,
if it’s not documented, it’s highly unlikely we are going to remember it
’til the next time. Coding that includes
all these key medical concepts is also important. Coding professionals and others
have gone through a lot of training
to look at the documentation and come up
with the proper code, but if the documentation
isn’t there, they simply can’t
come up with a code to represent accurately what
that patient’s condition is. So if we look at coding,
I think one of the things that becomes very apparent is
that we are looking at an era where we’re trying to get
a lot better documentation, resulting
in a lot better coding, resulting in a lot better data
for a whole variety of purposes. Accurate coding is a requirement
for good patient care; good care is important to improve the quality
of care, overall, the effectiveness of care,
and ensuring patient safety; and complete
and accurate documentation of important clinical concepts
of the patient condition is a requirement
for good patient care. Requirements for documentation
to support ICD-10 are consistent
with the documentation to support good patient care
and improve healthcare data. I think we’re going through
an era where we’re seeing the specificity and detail
and relevance of coding become increasingly
more important, getting at better data. We see this
in other industries. We would not allow
the level of specificity to be the ultimate level
of vagueness if we order a pair of shoes
or if we order dinner, or we would not expect our pilot
to tell us we’re flying to an unspecified location and
landing at an unspecified time. You know, we expect specificity
in almost every other industry and it’s that point in time
where we really need to start looking at getting
better documentation, better data, for better
patient care, moving forward. Thank you for participating
in this presentation. I hope you found
some value in this. I wanted to point out
that there are a lot of great resources
out there. CMS has a site
that they’ve put up that includes a lot
of different resources. These resources allow us
to pull down codes, to pull down guidelines, to
pull down a lot of information. All of this information
is for free. There is a website called
“Road to 10” which is specifically designed
for small provider practices, to give some guidance,
in terms of how to go through this transition,
to identify key resources, to point to a number
of different resources. I think it’s important
to understand this and start looking
at this soon because the time
is closing quickly and it’s important
to get that understanding of how is this going to
impact your practice and how are you going
to best position yourself to get the most
out of this. So thanks again
for participating. You can now take
the CME/CE posttest by clicking on the
“Earn CME/CE Credit” link. Please also take a moment
to complete the program evaluation
that follows. Thank you.

6 Replies to “ICD-10 and Clinical Documentation”

Leave a Reply

Your email address will not be published. Required fields are marked *