Information Technology Reference
In-Depth Information
timeliness. Unlike the Charlson Index, not all patients can be classified using the APRDRG index.
The National Inpatient Sample contains several risk adjustment coding methods, including the
APRDRG severity and mortality indices. We will compare these indices to the Charlson Index discussed
in Chapter 5. In addition, we will try to determine through data mining whether some codes are included
or not included in computing the APRDRG indices. We will show how providers are ranked using the
APRDRG index to compute the expected mortality rank.
Background
Diagnosis related groups (DRG) are used throughout healthcare. Providers are paid a set fee for a patient
treatment. However, they are limited in that they do not distinguish between patient conditions; sicker
patients require more resources and health costs, but reimbursements are the same as they are for healthier
patients. The DRGs were originally implemented by Medicare for billing purposes. However, relying
only on the DRG without any reference to patient condition results in underpayments to those providers
who treat patients with the most severe conditions.(Antioch, Ellis et al. 2007) Problems with DRG in
relationship to patient condition were identified early on.(McNeil, Kominski et al. 1988) DRGs are not
accurate predictors of either costs or outcomes.(Gross, et al. 1988) One of the reasons is that complica-
tions resulting from care need to be distinguished from pre-existing conditions, and DRG codes (and
ICD9 codes) do a poor job of this.(Young, Macioce et al. 1990; Naessens and Huschka 2004)
One attempt to identify patient risk and identify those who need more resources was the development
of the refined DRG codes (RDRG). The RDRG codes subdivide the DRGs into levels of complexity that
describe the patient condition. There are four general classes for RDRG refinement: no comorbid condi-
tions, moderate comorbid conditions, major comorbid conditions, and catastrophic comorbid conditions,
although only surgical patients qualify for the catastrophic class.(Leary, Leary et al. 1993) Therefore,
there are four severity classes for surgical patients, but only three for medical patients. Clearly then,
a higher proportion of surgical to medical patients will result in a higher proportion of patients in the
highest category. A number of analyses have been performed to validate the RDRG, although problems
have occurred with it as well.(Cerrito 2007) Another system developed by Medicare (MS-DRG) uses
three tiers of severity to assign payments for specific DRGs.(Maurici and Rosati 2007) There are also
a number of attempts to define severity adjustment for specific groups of patients.(Barbash, Safran et
al. 1987)
The all patient refined diagnostic related group (APRDRG) is another modification of the DRG with
four classes of illness severity and four classes of mortality risk.(Pilotto, Scarcelli et al. 2005) Once the
APRDRG is defined, a hospital case mix is defined as the sum of all relative weights divided by the
number of Medicare cases where the weight is assigned based upon resource consumption in terms of
diagnostics, therapeutics, bed services, and length of stay. The severity and mortality indices are defined
as mild, moderate, severe, and extreme.(Lagman, Walsh et al. 2007) Thus, there is the base APRDRG
value for the primary reason the patient is in the hospital along with the severity of illness subclass and
the risk of mortality subclass. These subclasses are very specific to the value of the base DRG. Thus,
patients with different DRG codes cannot be compared directly.
Details about the index are posted on the HCUP website (Healthcare Cost and Utilization Project) at
http://www.hcup-us.ahrq.gov/db/nation/nis/APR-DRGsV20MethodologyOverviewandBibliography.pdf.
The first step to defining the APRDRG values is to eliminate any and all secondary diagnoses that are
Search WWH ::
Custom Search