 |
Claims  |
|
|
It is claimed:
1. A method of monitoring the performance of a health provider in rendering
medical services to a patient, comprising the steps of:
(1) predicting an outcome of treatment of the patient by the health
provider by a method including the following steps executed on a computer:
converting into digital electronic signals input medical codes of at least
a patient's principal diagnosis, any secondary diagnoses, any surgical
procedures, sex and age,
identifying a few of an existing plurality of diagnostic and/or procedure
groups of illnesses that are appropriate for either one of the surgical
procedures signals or the principal diagnosis signals,
selecting from those few identified groups a single diagnostic or procedure
group that is consistent with the signals of any secondary diagnoses and
the patient's age and sex, and
calculating a sub-category within the single selected diagnostic or
procedure group for predicting at least one outcome by a formula that
mathematically combines various of the medical code input signals with
constants unique to the single selected diagnostic or procedure group, the
constants having been determined from a statistical analysis of a large
amount of actual patient data within the single selected diagnostic or
procedure group by use of the same formula;
(2) predicting from the calculated sub-category at least one outcome of
patient treatment;
(3) comparing said predicted outcome with the actual outcome of patient
treatment;
(4) monitoring the performance of health providers through use of said
comparison; and
(5) providing counseling to said health provider if the performance level
of the health provider falls below an established level.
2. A method of monitoring the performance of a health provider in rendering
medical services to a patient, comprising the steps of:
(1) in a computer system having data of a medical patient that includes the
patient's age, a numerical representation of the patient's sex, an ICD
code of a principal diagnosis of the patient's condition, one or more ICD
codes of secondary diagnosis of the patient's condition, and a DRG
determined from the ICD codes in accordance with a Government grouper
algorithm, a method of predicting by computer processing an outcome of
treatment of the patient, comprising the steps of:
providing a computer static data base that includes a first table of DRG
categories into which each of a plurality of ICD codes is mapped, and a
second table of Government mandated weights of the individual DRG
categories,
determining from the first table the DRG categories into which each of the
ICD codes is mapped,
reading from the second table the weights of each of the determined DRG
categories,
summing the read DRG weights, and
predicting an outcome of the patient treatment by use of the sum of the
read DRG weights;
(2) comparing said predicted outcome with the actual outcome of patient
treatment;
(3) monitoring the performance of health providers through use of said
comparison; and
(4) providing counseling to said health provider if the performance level
of the health provider falls below an established level.
3. A method according to claim 2 wherein the outcome predicting step
includes estimating the length of hospital stay for the patient and
classifying that stay length into one of a few categories within the DRG
determined for the patent.
4. A method according to claim 2 wherein the method of predicting an
outcome of treatment additionally comprises a step of counting the number
of patient ICD diagnosis codes, and wherein the outcome calculation step
includes use of said ICD code count in estimating the outcome.
5. A method according to claim 4 wherein, in the method of predicting an
outcome of treatment, the computer system additionally includes one or
more ICD codes of the procedures performed or to be performed on the
patient, and wherein the outcome predicting method additionally comprises
a step of counting the number of patient ICD procedure codes, and further
wherein the outcome calculation step includes use of said ICD code count
in predicting the outcome.
6. A method according to claim 5 wherein, in the method of predicting an
outcome of treatment, the first table of the static data base also
includes a MDC body system category into which each ICD code is mapped,
and wherein the outcome estimating method additionally comprises a step of
counting the number of MDC categories in which the patient ICD codes map,
and further wherein the outcome calculation step includes the use of said
MDC category count in predicting the outcome.
7. A method according to claim 2 wherein, in the method of predicting the
outcome of treatment,
the first table of the static data base also includes an indication for the
individual ICD codes whether each is on a Government list of ICD codes
that can indicate the existence of a complication or co-morbidity, and, if
so, also indicates a weight of severity of the condition indicated by each
such ICD code,
said method additionally comprising the steps of reading the weight of the
patient's ICD codes and summing those weights, and
further wherein the outcome calculation step includes the use of the sum of
said weights in estimating the outcome.
8. A method according to claim 2 wherein, in the method of predicting the
outcome of treatment,
the second table of the static data base also includes for individual DRGs
one or more DRGs that are medically related thereto and which have higher
Government assigned weights,
said method additionally comprising the steps of reading from the second
table the related DRGs for each DRG into which a patient ICD code is
mapped, and looking up in the first table the Government assigned weights
for each of the related DRGs so read, and summing those read weights, and
further wherein the outcome prediction step includes the use of related DRG
weight sum in predicting the outcome.
9. A method of monitoring the performance of a health provider in rendering
medical services to a patient, comprising the steps of:
(1) for use with a system that estimates the amount of resources that are
likely to be consumed to treat a specific patient illness by classifying
the patient into one of a large number of categories of estimated resource
consumption on the basis of input information that includes a principal
diagnosis and/or a surgical procedure performed or to be performed, any
secondary diagnoses, the patient's age and sex, a method of more
accurately estimating an outcome of treatment of the patient by a health
provider within said one category, comprising the following steps executed
on a computer:
determining a category for each of a plurality of diagnoses and/or
procedures of the input information as if it was the only diagnosis or
procedure to be considered, and noting the resource estimate for each such
category,
determining from said input information a plurality of other quantities
relating to the patient's condition, and
calculating a refined resource estimate within the determined one category
by solving an algebraic equation that combines the noted category resource
estimates and the other quantities derived from the input information
after multiplication by a plurality of constants, said constants having
been derived by statistical analysis of a large amount of actual data of
input information and the resources consumed by patients classified in
said one category by use of said algebraic equation;
(2) comparing said calculated resource estimate with the amount of
resources actually used;
(3) monitoring the performance of health providers through use of said
comparison; and
(4) providing counseling to said health provider if the performance level
of the health provider falls below an established level.
10. The method according to claim 9 wherein the step of determining a
category includes doing so for each of the diagnoses in said input
information but not for any surgical procedures therein.
11. The method according to claim 9 wherein the step of determining a
category includes doing so for each of the principal and secondary
diagnoses and surgical procedures in said input information.
12. The method according to claim 9 wherein the step of determining other
quantities includes calculating by said computer the following from the
input information:
a total of the number of different diagnoses,
a total number of any surgical procedures performed or to be performed, and
a number of different body systems involved as a result of the patient's
illness.
13. A method of monitoring the performance of a health provider in
rendering medical services to a patient, comprising the steps of:
(1) for use with a system that estimates the amount of resources that are
likely to be consumed to treat a specific patient illness by classifying
the patient into one of a large number of categories of estimated resource
consumption on the basis of input information that includes a principal
diagnosis and/or a surgical procedure performed or to be performed, any
secondary diagnoses, the patient's age and sex, a method of forecasting an
outcome of treatment of the patient by a health provider, comprising the
following steps executed on a computer:
determining a plurality of quantities relating to the patient's illness,
such a determination being made only from (A) the input information about
the patient and the patient's illness that is used to determine said one
category, and (B) a static database including but not limited to
information about the categories, diagnoses, procedures and category
estimated resource consumption, and
calculating a refined resource estimate within the determined one category
by solving an algebraic equation that combines said plurality of
quantities after multiplication by a plurality of constants, said
constants having been derived from use of said algebraic equation by
statistical analysis of a large amount of actual data of input information
and the resources consumed by patients classified in said one category;
(2) comparing said calculated resource estimate with the amount of
resources actually used;
(3) monitoring the performance of health providers through use of said
comparison; and
(4) providing counseling to said health provider if the performance level
of the health provider falls below an established level.
14. The method according to any of claims 9-13, inclusive, wherein said
diagnoses and surgical procedures are expressed as ICD codes, said
categories are diagnostic related groups (DRGs) of a health care
reimbursement system, and the category resource estimates are Government
designated weights of the individual DRGs.
15. A method of monitoring the performance of a health provider in
rendering medical services to a given patient from patient information
including illness diagnoses, surgical procedures performed or to be
performed, if any, age and sex, comprising the steps of:
establishing on a computer a mathematical relationship of said treatment
outcome with a plurality of variables developed from said patient
information,
inputing to the computer data including said information and resulting
treatment outcomes for a large population of actual patients,
utilizing the mathematical relationship to determine by calculations with
the computer a set of elements in the mathematical relationship from said
actual patient data which minimize differences between the actual outcomes
and those calculated by said mathematical relationship from the actual
patient information,
forming by calculation on the computer the mathematical relationship
variables from said given patient information,
solving on the computer said mathematical relationship with said given
patient variables and said determined set of elements, thereby to provide
a prediction of the outcome of treatment for said given patient;
comparing said predicted outcome with the actual outcome of patient
treatment;
monitoring the performance of health providers through use of said
comparison; and
providing counseling to said health provider if the performance level of
the health provider falls below an established level.
16. A method according to claim 15 wherein the outcome of treatment
includes the level of resources expended in providing patient treatment.
17. A method according to claim 15 wherein the step of forming the
variables includes the steps of forming from said patient information at
least a first variable that is proportional to the total number of said
diagnoses, and a second variable that is proportional to the total number
of said surgical procedures performed or to be performed.
18. A method according to claim 15 wherein the step of forming the
variables includes the step of forming a variable that is proportional to
a sum of relative weights of groups of the diagnoses and/or procedures.
19. A method according to claim 15 wherein the step of establishing a
mathematical relationship includes the steps of establishing a plurality
of constants as said elements, and obtaining mathematical products of one
or more of said variables by one of said constants, and then summing the
products in order to obtain the outcome.
20. A method according to any one of claims 15-19 wherein the step of
forming the mathematical relationship variables includes doing so only
from patient information of illness diagnoses, surgical procedures
performed or to be performed, if any, age and sex.
21. A method of monitoring the performance of a health provider in
rendering medical services to a patient from patient information including
illness diagnoses, surgical procedures performed or to be performed, if
any, age and sex, comprising the steps of:
(1) in a computer system that includes means for mapping a medical
patient's condition and/or treatment into a single group of related
diagnoses and/or procedures from information of the patient that includes
a principal diagnosis code, one or more surgical procedure codes, if any,
one or more secondary diagnosis codes, if any, the patient's age, and the
patient's sex, a method of predicting an outcome of treatment of the
patient by a health provider by use of said computer system, comprising
the steps of;
calculating a plurality of variables from said patient information, and
mathematically combining said variables with the use of a plurality of
constants that have been determined by statistical analysis of data from
an actual patient population within said single group of related diagnoses
and/or procedures, thereby to predict an outcome of treatment of the
patient;
(2) comparing said predicted outcome with the actual outcome of patient
treatment;
(3) monitoring the performance of health providers through use of said
comparison; and
(4) providing counseling to said health provider if the performance level
of the health provider falls below an established level.
22. A method according to claim 21 wherein the outcome of treatment
includes an indication of the level of resources expended in providing
patient treatment.
23. A method according to claim 21 wherein the step of calculating a
plurality of variables includes the steps of forming from said patient
information at least a first variable that is proportional to the total
number of said diagnoses, and a second variable that is proportional to
the total number of said surgical procedures performed or to be performed,
if any.
24. A method according to claim 21 wherein the step of calculating a
plurality of variables includes the step of forming a variable that is
proportional to a sum of relative weights of various groups of related
diagnoses and/or procedures into which the diagnoses and/or procedures
individually map.
25. A method according to claim 21 wherein the step of mathematically
combining the variables includes obtaining a plurality of mathematical
products of one or more of said variables by one of said constants, and
then summing the products in order to obtain the outcome.
26. A method according to claim 21 wherein the steps of predicting of the
outcome of treatment comprise the additional step of classifying the
predicted outcome within one of a plurality of sub-categories within the
determined single group of related diagnoses and/or procedures.
27. A method according to any one of claims 21-26 wherein the step of
forming the mathematical relationship variables includes doing so only
from patient information of illness diagnoses, surgical procedures
performed or to be performed, if any, age and sex. |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
BACKGROUND OF THE INVENTION
This invention relates generally to the identification of quality and cost
efficient medical providers, and specifically to computer software
techniques and systems for estimating what the cost to treat a patient
should be, based upon the condition of the patient and to the extent that
any treatments or procedures impact the patient's health status.
Due to the geometric escalation of medical care costs, there is increased
pressure on public policy makers to establish cost containment programs.
For this reason state and Federal governments are beginning to adopt
various case specific or case-mix reimbursement systems. The Social
Security Amendments of 1983, (Public Law 98-21), introduced a diagnosis
specific prospective payment system that has been incorporated into the
Medicare reimbursement policies. Under this system, the amount of payment
for a patient hospital stay is determined by a one of hundreds of
government defined Diagnostic Related Groups ("DRGs") into which the
patient stay is categorized based upon diagnoses and procedures performed.
Hospitals are reimbursed according to a fixed schedule without regard to
actual costs to the hospital in rendering medical services to the patient.
It is expected that this same reimbursement policy will in time be
extended to establish the level of reimbursement to other health care
providers and/or from other government entities and insurers.
The DRGs represent a statistical, clinical classification effort to group
together those diagnoses and procedures which are clinically related and
have similar resource consumption. A DRG that is appropriate for a given
hospital stay is selected, under the reimbursement system, by a particular
set of patient attributes which include a principal illness diagnosis,
specific secondary diagnoses, procedures performed, age, sex and discharge
status (i.e., how the patient left the hospital, whether the patient was
transferred, died, etc.). The principal diagnosis is that which caused the
patient to be hospitalized, even though the patient may have even more
serious problems, as would be indicated by secondary diagnoses. If a
surgical procedure is performed, the DRG is determined primarily by that
procedure. If no procedure is performed, the DRG is determined primarily
by the principal diagnosis. The treatment of a patient during a single
hospital stay is classified in only one DRG.
As shown in the few examples of DRGs given in Table I attached hereto, a
fixed reimbursement factor (relative weight) is assigned to each DRG by
the government. This determines the amount the hospital will be reimbursed
for treatment of a patient that falls within the DRG, regardless of the
hospital's cost or what the charges would have been for a non-Medicare
patient. The more complex diagnoses or procedures that typically consume
more resources should result in a higher paying DRG. (See the different
relative weights in the example DRG's of Table I.) In addition to the
reimbursement factor, each DRG has an average length of stay (LOS) in days
assigned as another measure of the consumption of resources that is
expected to treat a patient whose attributes cause that particular DRG to
be selected.
There are currently 473 DRGs which cover all patients treated under
inpatient conditions. These are set forth in the regulations of the Health
Care Financing Administration. The example DRGs of Table I attached hereto
are taken from those regulations. Since adoption of the system,
regulations have been issued annually that make some changes in
classification details to take into account experience under the system.
Under the current version of this reimbursement system, the hospital does
not directly determine the appropriate DRG category for services rendered
a Medicare patient. Rather, the hospital submits an appropriate Federal
form (currently form UB-82) after discharge of the patient, which includes
codes from a standard coding system to identify the primary and secondary
diagnoses made, and any procedures performed, and gives patient
information that is relevant to determining the appropriate DRG category,
such as age and sex. As an alternate to using such a form, the coded
information can be submitted on magnetic media, such as tape, in computer
readable form. From this information, an intermediary reimbursing agent,
or the Health Care Finance Administration itself, determines the proper
DRG, and thus the amount of reimbursement.
The commonly used notation "ICD-9-CM" means the International
Classification of Diseases--9th Revision, Clinical Modification, and
refers to a coding system based on and compatible with the original
international version of the ICD-9 coding system provided by the World
Health Organization. The ICD-9-CM coding system is used in North America,
and it is a classification of diseases, injuries, impairments, symptoms,
medical procedures and causes of death. These codes are listed in detail
in a publication of the Commission on Professional and Hospital
Activities, Ann Arbor, Michigan, entitled "ICD-9-CM", Jan. 1, 1979. It is
likely that the classification system will be revised and a 10th revision
forthcoming within a few years. The techniques being described herein are
not limited to a particular version of the ICD diagnosis and procedure
classification system but rather will use whatever system is current at
the time. As a shorthand reference to that system, the term "ICD" will be
used hereinafter, unless a specific version is being discussed as an
example.
The ICD coding system was designed for the classification of morbidity and
mortality information for statistical purposes and for the indexing of
hospital records by disease and operations for data storage and retrieval.
The ICD codes are initially divided into Disease and Procedure sections.
These sections are further divided into subsections which encompass
anywhere from 1-999 three digit disease or 1-99 two digit procedure code
categories. Within the three digit code categories there can be an
additional 1 or 2 decimal digits to divide the codes into subcategories
which further define the disease manifestations and/or diagnostic
procedures. There are approximately 15,000 ICD codes. Only a portion of
these are relevant for determining Medicare payments. The DRG Medicare
payment system first involves the coding of diagnostic and procedural
information into ICD code numbers by hospital medical records clerks
before a patient can be assigned a DRG.
Each DRG is determined in part by an ICD code for the principal diagnosis,
and ICD codes for each procedure that may have been performed. There are
also ICD codes for identifying complications occurring during treatment,
and the existence of any co-morbidities (i.e., secondary diagnoses of
conditions other than the principal disease existing at the time of
admission). These, as well as the patient's age, sex, and discharge
status, determine a particular DRG for the patient.
It is possible that a large number of sets of ICD numbers or codes can lead
to the same DRG. Table II attached hereto lists the ICD-9-CM codes that
currently fall within each of a few of the DRGs that are used as examples
in Table I. The existence of any one operative surgical procedure ICD-9-CM
code listed under DRG 261, for example, will cause that DRG to be
selected. The remaining DRG examples of Table II (numbers 31, 32 and 33)
each have the same list of ICD-9-CM codes that will cause a DRG to be
selected. A single one of these related DRGs is then selected based upon
the age of the patient and whether there exists any complication or
co-morbidity (referred to together as a "C.C." in the DRG definitions) as
evidenced by an appropriate secondary diagnosis ICD-9-CM code.
A patient's age is a part of the definition of many of the DRG categories,
as shown by the examples of Tables I and II. Pediatric patients (age 17 or
less) and elderly patients (age 70 years or more) often fall into separate
DRG categories that otherwise have the same textual definition. Where this
occurs, the hospital is paid more for treatment of the older patient by
assigning a higher paying DRG.
Also, the presence of a complication or comorbidity (C.C.) with a patient
is a part of the definition of many DRGs. A patient with a complication or
comorbidity is considered to be a sicker person for certain illnesses than
one without a C.C. and the hospital is reimbursed more for those illnesses
by classifying such a patient in a higher paying DRG. However, not all
medically recognized complications or comorbidities are recognized by the
DRG reimbursement system to have any effect on the payment to be made. The
DRG Medicare system currently specifies about 3000 of the approximately
15,000 ICD-9-CM codes as effective to establish a C.C. and thus provide
higher reimbursement. That is, if any one of these approximately 3000
ICD-9-CM codes appear as a secondary diagnosis on the patient discharge
information, a C.C. is deemed to exist. Thus, if the patient is otherwise
classified into a family of DRGs where the existence of a C.C. makes a
difference, the higher paying DRG is selected from the family. For
example, with regard to the family of DRGs 31, 32 and 33 of Tables I and
II which have the same lists of principal diagnoses ICD-9-CM codes, both
the patients age and the existence or non-existence of a C.C. determines
which of the three DRG categories is selected for reimbursement purposes.
It should be noted that the relative weight and mean length of stay (LOS)
definitions of those three DRGs vary widely.
The actual reimbursement that a hospital receives for each patient involves
the multiplication of the relative weight of the DRG (see Table I) with
other factors set by the Federal government. These other factors are
determined by statistical variables (e.g. cost data of that particular
hospital for a period, the type of patients a hospital treats in relation
to the hospital's resources expended for those patients, and the wage and
cost of living index).
Computer software is available for calculating the appropriate DRG from the
input codes that are provided by the hospital. A DRG Grouper System
converts the ICD codes of a patient's stay, along with the other DRG
related factors (age, sex, discharge status), are mapped into the
corresponding DRG category. This is public information that is available
at cost. One company manufacturing an enhanced DRG Grouper is the DRG
Support Group, Ltd., a subsidiary of Health Systems International, Inc.
Because such a large proportion of hospital patients fall under the
Medicare system (40% or more of the patients of some hospitals), the DRG
system is extensively used. It is natural, therefore, that the system
would also be used for health care management and evaluation purposes.
However, it is widely recognized that the grouping of patients resulting
from use of the DRG system does not have as high a degree of homogeneity
as is statistically desirable. That is, the resource consumption of a
population of patients, who all are classified into a single DRG, varies
widely. The statistical deviation from the single mean length of stay
(LOS) for most of the DRGs is large, apparently because the overall level
of sickness of the patients so grouped varies widely. The sicker patients
require more hospital resources to be devoted to them but the DRG Medicare
system considers this only to a limited extent by selecting a DRG
primarily from only a single principal diagnosis made or surgical
procedure performed.
As a result, there have been many suggestions for refining the DRG system,
or to go to a different system, in order to result in a more homogeneous
grouping of patients. The reasons for doing so include the need to have
data for monitoring hospital and physician performance, as well as
improving the reimbursement system itself. The suggested approaches
include many different ways to measure how sick a patient really is.
The primary variable in any patient population which must be taken into
account before either mortality and morbidity rates or resource
consumption can be addressed is that of the patient diagnosis. If a
physician is asked to predict the mortality rate of a group of patients,
the first question he or she will ask is "What is the diagnosis?". The
expected mortality rates between a fractured wrist and cerebrovascular
accident (stroke) are very different. The DRGs are a very adequate way of
subdividing the patients on the basis of their diagnoses.
The second and crucial variable is that of the severity of the patient's
illness within each of the various diagnoses. An example is that some
myocardial infarctions (heart attacks) are fatal, and some go completely
unnoticed by the patient, representing a wide variation in severity. The
DRGs do not have the ability to adequately determine the acuity (severity)
of the patient's illness within the diagnostic categories.
It is a primary object of this invention to provide a computer based
technique and system for estimating the severity of patients' illnesses
from hospital discharge data and other medical information, and for
estimating the resources likely to be consumed in the course of providing
medical service to patients, all with improved accuracy and convenience.
SUMMARY OF THE INVENTION
This and additional objects are accomplished by the present invention
wherein, briefly and generally, in a specific form, a computer system is
provided for calculating the severity of patients' illnesses and thereby
providing an estimate of resource consumption from the same information
that is used as a basis for determining the DRG. This has a significant
advantage in that the necessary data is readily available in a form to be
directly used in making the estimate, often stored on computer magnetic
media so that it can be fed directly into a computer making the estimate,
perhaps even the same computer that is determining the appropriate DRG. In
a preferred embodiment of the invention, no further information about the
patient and his or her condition is required. Hidden clinical information
is extracted by the resource estimating system from the ICD codes and
other available input data in order to make an estimate. This input data
is combined by the computer according to a formula (a linear equation, in
a preferred embodiment) that includes a set of constants that exist in a
static data base for each DRG included in the system. Variables of the
formula can include, for example, the number of different ICD codes
(particularly any secondary diagnosis codes) specified by the health
provider to describe the condition of patient during a particular
hospitalization, a sum of the Government weights for each of the DRGs into
which the ICD codes (particularly the diagnosis codes) are mapped, and a
sum of relative weights assigned to certain ICD codes that, when specified
as secondary diagnosis, indicate a sicker patient than is communicated by
the principle diagnosis code alone.
A set of such constants is determined for a given DRG by a process of
statistically analyzing a set of actual patient data for that DRG by use
of the same formula. The variables of the formula, including outcome,
specific diagnoses codes, etc., are in this case known for each actual
patient. In effect, a separate formula is established for each such
patient record. A statistical computer program then determines a set of
constants for use in the formula for the given DRG which minimizes
variances between the actual known outcomes and those estimated by use of
the formula.
In a preferred form, the estimated outcome is expressed in one of several
categories (such as categories 1 through 5) for the DRG that has been
determined for the patient, and represents a much more homogeneous
grouping of patients than is provided by the DRG categories themselves,
because it is based upon various levels of illness severity within each
diagnosis. Because physician and hospital providers can then be compared
on the basis of a homogeneous patient population, it becomes possible to
identify those providers whose practice patterns are of the highest
quality and most cost efficient.
There are two primary applications for such a system other than to
objectively calculate the amount of payment to the health provider. One is
a retrospective: the cost performance of a physician, hospital, or expense
of an insurance plan, for example, can be compared to the estimate for
one, a few, or a large number of patients. That is, a set of actual costs
incurred are compared with the estimates. This is quite a useful tool
since the estimates are made by formulas whose constants have been
calculated from a large amount of actual patient data.
The other primary application is concurrent: the expected cost of treating
a patient may be determined upon admission of the patient to the hospital
from the initial diagnosis and expected procedures. Any problem cases
identified by such estimates can then be managed more intensively than
others. Further, any changes occurring during the hospital stay, primarily
the addition or change of any of the secondary diagnosis data, can
evidence missed or erroneous initial diagnoses or complications caused by
disease progression or improper treatment, thus providing a tool for
monitoring hospital and/or physician performance.
Additional objectives, features and advantages of the various aspects of
the present invention will become apparent from the following description
of its preferred embodiment, which description should be read in
conjunction with the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a curve that illustrates the distribution of costs in treating
patients who all are classified into the same DRG;
FIG. 2 is a curve that illustrates the further classification that results
from the techniques of the present invention;
FIG. 3 shows a block diagram of a typical computer system that may be
utilized to practice the present invention;
FIG. 4 is a flow diagram that shows the general operation of public domain
software that determines the appropriate DRG for the purpose of
reimbursing a health provider for services to a particular patient;
FIG. 5 is a flow diagram of a computer program which operates w the
computer system of FIG. 3 in carrying out various aspects of the present
invention, according to a first example;
FIGS. 6-8 provide a flow diagram of a computer program which operates with
the computer system of FIG. 3 in carrying out various aspects of the
present invention, according to a second example; and
FIGS. 9-11 illustrate a portion of the process for calculating the value of
constants used in the computer program examples of FIGS. 5-8, according to
another aspect of the present invention.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
General Discussion of the Invention
The non-severity adjusted, and therefore non-homogeneous grouping of
patients, that results in many Diagnostic Related Groups (DRGs) of the DRG
Medicare reimbursement system is illustrated in FIG. 1. The number of
patients classified in a single DRG is plotted against their actual
resource consumption incurred in treatment, in terms of charges or mean
length of stay (LOS) of the patient in the hospital. The fact that the
distribution deviates significantly from the single mean LOS used in the
DRG system as a basis for payments shows the need to modify the system
when a high degree of certainty in estimates of resources consumed is
required.
Therefore, according to the present invention, a sub-category of resource
consumption is calculated in addition to determining a DRG by the usual
method. This sub-category, referred to herein as the "acuity index", is
calculated from the same information used to determine the DRG, as
described in detail later in this description. Hidden clinical information
of the overall level of sickness of a patient, not now used by the DRG
system, is extracted from input data common to the DRG system. By using
the same input data, a simplicity in implementation results. By expressing
the result of an acuity index within the DRG system, there is conformity
with a system familiar to health care professionals.
FIG. 2 shows an idealized plot of actual resource consumption for a given
patient population in a single DRG versus resource consumption that is
estimated (predicted) by the techniques of the present invention. Rather
than give as a result the estimated LOS or other quantitative indication
of resource consumption, the estimated resources are expressed as falling
within one of five acuity indices for that particular DRG. Acuity index
number 1 includes the patients requiring the lowest amount of resources
within the given DRG. Acuity index number 5 includes the patients with the
highest estimated resource consumption. Acuity index number 3 generally
includes the government calculated mean LOS within it. Of course, the
exact number of sub-groups selected for use in any system depends upon its
specific application.
Before describing the method of determining the acuity index for any given
patient, a computer system used to calculate both the DRG and acuity index
for a patient is outlined. Referring to FIG. 3, the main components of a
computer system that is suitable for implementing the various aspects of
the present invention is shown. Connected to a common computer bus 11, are
several operational units that form the computer system. These are a
central processing unit (CPU) 13, a main R.A.M. memory 15, a disc drive
17, a printer 19, and an entry keyboard and CRT terminal 21. A second such
terminal 23, and perhaps even additional terminals, can be provided as
desired. A modem 25 is optionally provided to add a communication
capability.
A principal magnetic disc data file that is part of the system of FIG. 1 is
that containing information of all the DRGs. It is accessible by the disc
drive 17. The DRG file is a static computer database having one record for
each DRG number. Table I shows seven fields of information for each of the
example DRGs given. This information is that published as part of the
Federal regulations. The first stored item of information is the DRG
number, some unique number between 1 and 473. The next item of
information, shown in the second column of Table I, is the Major
Diagnostic Category (MDC) in which the individual DRG falls. The 473
specific DRGs are grouped by the Federal regulations into 23 MDCs of
related DRGs. Each MDC is defined to include the DRG's directed to matters
of a different body system than the others
The third column of Table I identifies the sex of the patient for which the
individual DRGs are appropriate. A "B" means the DRG is appropriate for
both sexes, a "F" for female only, and a "M" for male only. This field is
compared with the sex of the patient to check for an erroneous DRG
determination.
The fourth item of information for each DRG maintained in the static DRG
database is shown in the third column of Table I, namely the title or
textual description of the DRG. What is shown in Table I are all of the
items published in the Federal Register. It will be noticed that the
textual descriptions are very brief, so it may be desirable to expand that
description into medically relevant terminology for the purpose of the DRG
database herein.
The fifth data field, shown in the fifth column of Table I, is a relative
weight for each DRG. As discussed above, this determines the amount of
compensation that is given a hospital for treating a patient whose
diagnosis and/or procedures cause a particular DRG to be designated.
The same is true for each of the last two items of Table I. The column
"Mean LOS" is a calculated average length of stay for a patient within the
DRG. The "Outlier Cutoffs" column carries information of a maximum length
of stay in days that should be allowed. These last two items of
information are desirably printed whenever information as to its DRG is
printed in order to provide hospital staff with these guides, but they are
not otherwise used in the computer system to be described.
It will be noted from the titles in Table I that some of the DRGs include
an operative patient age range for their operation, and whether a C.C.
(complication or co-morbidity) is required for that DRG to apply. The
existence of a C.C. follows from a secondary diagnosis ICD code that is
one of about 3000 put on a special list by the current Federal system as
indicating a sicker patient that justifies higher compensation. Note from
the titles of the DRGs in Table I that DRGs 34 and 35 have the same titles
except for the reference to the patient's age and the existence of a C.C.
A patient with a principal diagnosis that indicates "other disorders of
the nervous system" will be classified into DRG 35 unless the patient is
either over 69 years of age or has a C.C. If the patient is older than 69
or has a C.C., then DRG 34 is appropriate. Note that the relative weight
and mean LOS of DRG 34 is greater than those of DRG 35. DRGs 34 and 35 are
referred to as a "doublet" since they result from a common principal
diagnosis; the appropriate DRG is then determined from the age and C.C.
data.
DRGs 96, 97 and 98 are a "triplet" since they all deal with bronchitis and
asthma, the particular DRG depending upon age and/or C.C. status. DRG 261
is given as an example of a single DRG that is applicable to a particular
diagnosis and/or procedure. The patients age or C.C. status is not
important for classification when such a DRG is the appropriate one.
The DRG table database is a static one and the same for all hospitals. For
the computer system to operate, certain information of the patient is also
required, primarily his or her age and sex. This information can be
entered directly when the computer system is being operated, or,
preferably, is maintained in a dynamic cur | | |
