Words: 4718
A Severity Adjustment Methodology Quality Management In Healthcare Abstract All Patient Refined Diagnostic Related Groups (APR-DRGs) currently represent one of the most widely used systems for severity adjustment of hospital outcome comparisons. Severity Adjustment is the method used to account for differences in patient characteristics such as age, severity of illness, and risk of mortality independent of the actual medical treatment given.
As a result, it’s possible to make better comparisons of hospital data and allow facilities to substantiate the notion that “my patients are sicker. ” Data sources for severity adjustment currently can be clinical (e. g. , a medical record) or administrative (e. g. , discharge abstracts). APR-DRGs are a refinement of Medicare’s DRG system and incorporate severity of illness and risk of mortality measures. Studies show APR-DRGs perform about as well, and in some cases better, than competing severity adjustment systems based on administrative data.
Overall, their widespread usage can be explained by their low cost, flexibility, and methodology. APR-DRGs: A Severity Adjustment Methodology Introduction All Patient Refined Diagnostic Related Groups (APR-DRGs) are a patient classification system developed by 3M Health Information Systems. They are a refinement of Medicare’s current DRG system to include severity of illness and risk of mortality measures. In APR-DRGs, severity of illness is defined as the extent of organ system loss of function or physiologic decompensation, while risk of mortality is the likelihood of dying (Averill, 2002).
According to the Centers for Medicare and Medicaid Services (CMS), APR-DRGs are one of the most widely used severity adjustment methodologies for comparative hospital performance (Medicare Program, 2006). Therefore, the purpose of this paper is to define what severity adjustment is, why it’s important, and the data sources available. An in-depth history and breakdown of APR-DRGs, who is using them, and an assessment of their use as a severity adjustment methodology will follow this. What is Severity Adjustment?
Severity adjustment is a method used to account for differences in patient characteristics (e. g. , age, income, and type of illness needing treatment) likely to affect the outcome of care (e. g. , death, physical functioning, resource utilization, and cost), independent of the actual medical treatment given. The purpose of severity adjustment (sometimes referred to as severity-of-illness adjustment) is to allow for a fair comparison of health outcomes, such as death or disability level (Iezzoni, 1997). Comparing health outcomes such as mortality and morbidity is one way to evaluate the quality of care.
Although, for example, we might like to compare mortality rates or functional status after a particular type of surgery, it is important to recognize that patients who die or recover more slowly after an operation may not have received poorer quality care but may have been sicker before treatment. Patient characteristics such as age, severity of illness (e. g. , localized cancer vs. metastatic cancer), and co-morbidities or secondary conditions (e. g. , the patient with cancer is also diabetic) place patients at different risks for outcomes before receiving care (Iezzoni et al. 1996). Severity adjustment is used to refer to adjustments made both to reflect the severity of the disease that is the focus of treatment and the overall illness level of the patient, including co-morbid conditions. Because both types of severity can contribute to the likelihood of achieving good outcomes, both are included in severity adjustment models. By accounting for these types of baseline patient characteristics or risk factors, severity adjustment enables comparisons of health outcomes to be made (Iezzoni, 1997). Why is Severity Adjustment Important?
In the absence of severity adjustment, comparisons of health outcomes are made that are often false and may lead to incorrect conclusions and actions. Hospitals that treat a greater number of higher risk patients than others should not be unfairly judged for accepting a riskier patient population. For example, a comparison of mortality rates between two long term acute care hospitals shows hospital A with a 30% rate and hospital B with a 10% rate. Without more information, hospital A appears to have worse mortality outcomes than hospital B.
However, hospital A specializes in treating elderly, medically complex respiratory patients requiring ventilator management and hospital B specializes in complex wound care treatment. Hospital A treats significantly sicker patients that are more likely to die compared to hospital B’s patient mix. The ability to adjust the mortality data by severity of illness would level the playing field in terms of comparing these hospitals and adjust the mortality rates by the severity of the patients they treat. Without severity adjustment, information may be inaccurate, misleading, or simply wrong.
Furthermore, incentives will exist for healthcare plans and medical groups to select the healthiest enrollees to improve performance (e. g. , to demonstrate that they have lower mortality and morbidity). More and more, information about health outcomes serves as the basis for policy decisions by providers, insurers, and employers. If there is an interest in comparing health outcomes, severity adjustment will be necessary (Iezzoni et al. , 1996). What Data Sources are Available for Severity Adjustment? Data sources used for severity adjustment affects the reliability and validity of the severity adjustment system.
There are currently two main data sources that contain information useful for severity adjustment. The first is clinical data, which can be found in the medical record, computerized laboratory results, and pharmacy records. The second is administrative data, which is information generated from an interaction between a patient and provider that mainly contains information about the costs, diagnoses, and services provided. Each of these data sources has advantages and disadvantages when it comes to their usefulness for severity adjustment data collection.
Clinical data taken directly from a patient’s medical record typically involves the ability to abstract information on things such as blood pressure, white blood count, and medications prescribed. Medical records contain information on a patient’s medical history, symptoms, physical examinations, diagnostic test results, response to treatment, discharge plans, clinical course, and demographic factors (McGlynn et al. , 1998). The information in medical records documents what occurred from a clinician’s perspective.
All of this information is confidential and special permission is required to gain access to any of the data contained in the medical record. The greatest strength of medical records is that they provide the detailed clinical data necessary for evaluating quality of care and is therefore extremely important when it comes to utilizing this data for severity adjustment. Clinical data represents the most important source of information about the details of diagnosis and treatment, patient risk factors, and the clinical outcomes of care. These data are required to examine differences in patient outcomes resulting from treatment (McGlynn et al. 1998). Disadvantages of obtaining clinical data from medical records are that there is typically no standard format or procedure for recording patient information. Records cannot be used for direct analysis due to the lack of uniformity, the information is often handwritten, and they tend to exist in hard copy format only. In addition, the abstract costs associated with clinical data are extremely high and time consuming when compared to administrative data (McGlynn et al. , 1998). Administrative data or information describing a hospital visit typically comes in the form of a discharge abstract.
Discharge abstracts are produced by hospitals on the Medicare claim forms (UB-92) for all hospitalizations in states that require them. Discharge abstracts include patient demographic data; payer information; principal and other diagnoses and procedures coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM); admission source; charge data; and discharge disposition (Iezzoni, 1997). Administrative data are typically computerized, which facilitates their use in analysis because large quantities of data are available.
In addition, the cost of capturing administrative data is typically low and data generally available immediately or within a few days (McGlynn et al. , 1998). Disadvantages involve the limitations on the capture of clinical information. UB-92 claim forms only allow the reporting of nine diagnosis codes and six procedure codes. Therefore, it’s possible that patients with diagnoses greater than nine do not have their clinical picture fully reported. Accuracy and specificity of diagnosis coding also affect the usefulness of administrative data for severity adjustment measures by potentially underestimating or overestimating clinical complexity.
Finally, patient outcomes (other than death) and functional status are not available from administrative data. Despite many of the identified disadvantages, APR-DRGs utilize administrative or discharge abstract data in assigning severity of illness and risk of mortality measures. The advantages of uniformity, widespread availability and computer readability make administrative data extremely useful for severity adjustment. How Did APR-DRGs Originate? APR-DRGs are derived from and based on the Diagnostic Related Group (DRG). DRGs are a patient classification scheme, which provides a means of relating the type of patients a hospital treats (i. . , its case mix) to the resources and costs incurred by the hospital (Averill, 2002). The design and development of the DRG began in the late sixties at Yale University as a means to categorize, classify, and describe all types of patient care (e. g. , Medicare, newborn, pediatric, and general populations) in acute care hospital settings (Bourcher, 2006). The first large-scale application of DRGs was in the late 1970’s when the New Jersey State Department of Health used DRGs for a prospective payment system, where hospitals were reimbursed a fixed amount for each patient (Boucher, 2006).
In 1983 Congress amended the Social Security Act to mandate a DRG-based hospital prospective payment system for Medicare patients (Boucher, 2006). There are currently three major versions of the DRG: basic DRGs, All Patient DRGs, and APR-DRGs. The basic DRGs are used by the Centers of Medicare and Medicaid Services (CMS) for hospital payment for Medicare Beneficiaries. The All Patient DRGs (AP-DRGs) are an expansion of the basic DRGs to be more representative of non-Medicare populations such as pediatric patients.
The APR-DRGs incorporate severity of illness and risk of mortality subclasses into the AP-DRGs. Since the APR-DRGs include both the CMS DRGs and the AP-DRGs, the development of all three versions of the DRGs will be reviewed. The history of how APR- DRGs came about is essential to understanding their use for severity adjustment. The DRG system was essentially designed by grouping or dividing all possible primary conditions for admission into mutually exclusive diagnosis categories referred to as Major Diagnostic
Categories (MDCs). The MDCs were formed by physician panels as the first step toward ensuring that DRGs would be clinically similar. Each MDC represents a single organ system or etiology that is related to a medical specialty area. Most MDCs correspond with major organ systems (e. g. , Respiratory System, Circulatory System, etc. ). Once MDCs were defined, each MDC were evaluated to look at the consumption of hospital resources and the presence or absence of surgical procedures performed in the operating room.
Each MDC was then divided into medical and surgical categories where medical DRGs were based on the patients’ principal diagnosis (reason for admission) and surgical DRGs were determined based on the surgical procedures performed. Since patients can have multiple procedures performed during a hospital stay, and a patient can be assigned to only one surgical group, the surgical groups in each MDC were defined in a hierarchical order. Patients with multiple procedures would be assigned to the surgical group highest in the hierarchy. In general specific groups of principal diagnoses were defined for medical patients.
For example, the medical groups or DRGs for the Respiratory System MDC are pulmonary embolism, infections, neoplasms, chest trauma, pleural effusion, pulmonary edema and respiratory failure, chronic obstructive pulmonary disease, simple pneumonia, RSV pneumonia and whooping cough, interstitial lung disease, pneumothorax, bronchitis and asthma, respiratory symptoms and other respiratory diagnoses (Hart, 2002). Once the medical and surgical groups were formed, each group of patients was evaluated to determine if complications, co-morbidities, or the patient’s age would affect the consumption of hospital resources.
Physician panels classified each diagnosis code based on whether the diagnosis, when present as a secondary condition, would be considered a substantial complication or co-morbidity (CC). A substantial complication or co-morbidity was defined as a condition, which because of its presence with a specific principal diagnosis, would cause an increase in length of stay by at least one day for at least 75 percent of the patients (3M Health Information Systems, 2003). Therefore, many DRGs were further divided based on the presence or absence of a CC condition.
This is the DRG system that was ultimately adopted by CMS in 1983 with the adoption of the Medicare prospective payment system (PPS). 3M Health Information Systems contracted with CMS to handle all revisions, definitions, logic, and documentation. The focus on all DRG modifications has been on problems relating to the elderly population. Pediatric and newborn populations were never really addressed by CMS modifications. The healthcare industry has utilized DRGs across a wide array of applications.
Hospitals have used DRGs as the basis of internal management systems. Medicaid programs and Blue Cross plans have used DRGS as the basis of payment systems. State data commissions have used DRGs as the basis for statewide comparative reporting systems (3M Health Information Systems, 2003). Therefore, the failure of the DRG update process to address problems for non-elderly populations became a serious limitation for most applications of the DRGs. The state of New York passed legislation in 1987, initiating a DRG prospective payment system for non-Medicare patients.
New York entered into an agreement with 3M Health Information Systems for DRG modifications and changes in DRGs and associated DRG software. The DRG definitions developed by New York and 3M are called All Patient DRGs (AP-DRGs). The AP-DRGs started to be used in New York State beginning January 1988 (3M Health Information Systems, 2003). AP-DRGs were developed for neonates, HIV and related diagnoses, trauma patients, long-term ventilator patients (tracheostomy), and transplant patients as well as many other conditions.
The AP-DRGs are also based in part on patient complications and co-morbidities and the effect these have on resources required for treatment (3M Health Information Systems, 2003). The AP-DRGs resulted in multiple changes being made to the CMS DRGs. Many changes were related to pediatric patient but others were related to all types of patients. Some of the AP-DRG changes have been adapted in the current CMS DRGs. The Centers for Medicare and Medicaid Services latter funded a project at Yale University on the use of CC’s in the DRGs (3M Health Information Systems, 2003).
The result of this research was the development of almost 1200 DRGs. The Yale research resulted in major improvement in the prediction of patient costs using composite data. However, there remained multiple concerns about the AP-DRGs, including the fact that patient severity of illness and risk of mortality were not predicted, many secondary diagnoses were not included in the system and there was concern about the effectiveness of the AP-DRGs and groupings of surgical and medical patients. However, the most important result was that the research demonstrated that co-morbidity subgroups could be created within the DRGs.
This was significant in terms of developing severity of illness categories (3M Health Information Systems, 2003). In developing the DRG and APR-DRG systems, the goal was to establish a relationship between the types of patients receiving treatment and the hospital resources that they used. However, as health care evolved and the use of post-acute care treatment increased, it became clear that there were limitations in the DRG system, which made it impossible to more completely evaluate patients in terms of severity of illness, risk of mortality and resource consumption used in their treatment.
These limitations led to the development of APR-DRGs by 3M Health Information Systems, which were designed to more completely describe the medical and surgical conditions of patients using factors (severity and risk of mortality) which were related to the treatment resources required (3M Health Information Systems, 2003). APR-DRG development involved a complex series of iterative patient classification assumptions, which were then compared to historical patient data and refining the assumptions as necessary, based on reliable clinical data.
The APR-DRGs are based on the medical and surgical categories in the CMS Major Diagnostic Categories (MDCs). The fundamental approach was to start with the MDCs for a patient and then assign an APR-DRG based on principal medical diagnosis (reason for admission) or operating room procedure. Although the basic APR-DRGs were used as a starting point, as the APR-DRGs developed, many changes and additions were made to refine the APR-DRG patient groups.
These changes and refinements have led to the current system which, for example, consolidates APR-DRGs based on complicated principal diagnoses, complicated operating room procedures, volumes of cases, difference of mortality, revised MDC definitions, and use of APR-DRG groups which clarify the principal diagnosis by considering multiple factors which cause a patients hospitalization (3M Health Information Systems, 2003). How Do APR-DRGs Generally Work? The APR-DRGs increase the specificity of the basic DRGs through adding four sub-classifications for every APR-DRG.
The sub-classifications describe patient’s severity of illness and risk of mortality. Severity of illness and risk of mortality are specific patient characteristics and therefore have separate sub-classifications. Taken together, the APR-DRG and severity of illness further refine CMS DRGs to provide more precise classifications of patients (3M Health Information Systems, 2003). The APR-DRG severity groups can be used to predict patient costs and lengths of stay.
The reason that the APR-DRGs are useful and effective is that patient severity of illness and risk of mortality are related to patient medical problems and patients with higher severity and risk of mortality often have multiple diseases and serious health problems that the APR-DRGs can represent. The APR-DRGs are focused on specific patient diseases and additional co-morbidity conditions that are related to the basic diseases, so taken together, this information can more precisely characterize the severity and risk of mortality patients.
How are Severity of Illness and Risk of Mortality Determined? In APR-DRGs, severity of illness and risk or mortality is primarily determined by the interaction of multiple diseases. Patients with diseases involving multiple organ systems are assigned to a higher severity level or risk of mortality subclasses because these patients are more difficult to treat and have poorer outcomes.
The process of determining the subclasses of severity of illness and risk of mortality subclasses for an APR-DRG begins by first assigning a severity of illness level and a risk of mortality level to each secondary diagnosis. The term “level” is used when referring to a categorization of a secondary diagnosis. The term “subclass” is used when referring to one of the subdivisions of an APR-DRG. For secondary diagnoses, there four distinct severity if illness levels and four distinct risk of mortality levels.
The four levels are numbered sequentially from one to four indicating, minor, moderate, major or extreme severity of illness or risk of mortality. Each secondary diagnosis is assigned to one of the four severity of illness levels and one of the four risks of mortality levels. The severity of illness level and risk of mortality level associated with a patient’s secondary diagnosis is just one factor in the determination of a patients overall severity of illness subclass and risk of mortality subclass (3M Health Information Systems, 2003).
The assignment of a patient to a severity of illness or risk of mortality subclass takes into consideration not only the level of the secondary diagnosis but also the interaction among secondary diagnoses, age, principal diagnosis, and the presence of certain OR procedures and non-OR procedures. The subdivision of each of the 314 APR-DRGs into four subclasses, combined with the two error APR-DRGs, which are not subdivided, results in 1258 APR-DRGs (3M Health Information Systems, 2003).
The process of determining the severity of illness or risk of mortality subclass of a patient consists of three phases. In phase I, the severity of illness level of each secondary diagnosis is determined. For example, a secondary diagnosis of uncomplicated diabetes is assigned a minor level; diabetes with renal manifestations would have a moderate level; diabetes with ketoacidosis would have a major level; and diabetes with hyperosmolar coma would receive an extreme level.
Once the level of each individual secondary diagnosis is established, then phase II determines a base subclass for the patient on all of the patient’s secondary diagnoses. In phase III, the final subclass for the patient is determined by incorporating the impact of principal diagnosis, age, OR procedures, non-OR procedure, multiple OR procedures, and combinations of categories of secondary diagnoses. The process is extremely complex with six steps to phase I, three steps to a phase II, and nine steps to phase III for a total of 18 steps (3M Health Information Systems, 2003).
There is a three-phase process of determining the risk of mortality subclass. This three-phase process parallels the three phases in the determination of the severity of illness subclass. In phase I, the risk of mortality of each secondary diagnosis is assigned a level of minor, moderate, major, or extreme. Once the risk of mortality level of each individual secondary diagnosis is established, then phase II determined a base risk of mortality subclass for the patient based on all the patients secondary diagnoses.
In phase III, the final subclass for the patient is determined by incorporating the impact of principal diagnosis, age, OR procedures, certain non-OR procedures, multiple OR procedures, and combinations of categories of secondary diagnoses (3M Health Information Systems, 2003). Table 1 below contains a simplified example of the assignment of a severity level for the same patient with four different clinical scenarios. The patient in these cases (starting with case 1) is a 50-year-old male with a long history of diverticulosis who is admitted for diverticulitis of the colon with an ulcer of the anus and rectum.
He was taken to the operating room where multiple affected areas of the large intestine were excised. Given the diagnoses and procedures of this patient, the APR-DRG assigned is 221 (Major small and large bowel procedures) with a corresponding severity level of one (minor) due to the fact there are no complications or co-morbidities. Cases 2, 3, and 4 demonstrate the effect on the severity level of adding additional complications or co-morbidities as secondary diagnoses. This is meant to demonstrate how APR-DRGs attempt to define a patient’s severity of illness given different clinical scenarios.
The resources, medical treatment, and risk of mortality are significantly lower for case 1 vs. case 4. While this is one patient example, it still represents the use of APR-DRGs to differentiate severity among patient populations. On a larger scale, the severity levels of all hospital patients can be used to “adjust” outcome measures so hospitals with mostly “case 4” patients are compared fairly with hospitals treating primarily “case 1” patients. Table 1 One Patient, Four Severity Levels |This table shows the clinical scenarios of the same patient given different secondary diagnoses for each case.
With each scenario, a higher| |severity of illness subclass is assigned. | |Principal Diagnosis: Diverticulitis of colon (562. 11) | |Procedure Performed: Multiple segmental resection of large intestine | | |Case 1 |Case 2 |Case 3 |Case 4 |Description | |Secondary Diagnoses |569. 41 |569. 41 |569. 41 |569. 1 |- Ulcer of anus & Rectum | | | |560. 9 |560. 9 |560. 9 |- Intestinal obstruction | | | | |422. 99 |422. 99 |- Acute myocarditis | | | | |426. 0 |426. 0 |- Atrioventricular block | | | | | |584. |- Acute renal failure | |APR-DRG |221 |221 |221 |221 |Major small and large bowel procedures | |Severity of Illness Level|Level 1 |Level 2 |Level 3 |Level 4 | | | |Minor |Moderate |Major |Extreme | | |APR-DRG Payment Rate |1. 4309 |1. 8794 |2. 492 |5. 9107 |The average of this “weight” for all patients | | | | | | |would determine a facilities “case mix” | Who is Using APR-DRGs? According to the Centers for Medicare and Medicaid Services (CMS) more than a third of the hospitals in the United States as well as 33 state agencies are using APR-DRG software as a severity adjustment methodology to analyze comparative hospital quality performance (Medicare Program, 2006).
The Medical Payment Advisory Committee (MedPAC) used APR-DRGs to analyze severity adjustments for the Medicare prospective payment system (PPS). In addition to being used in research by MedPAC, the APR-DRGs risk of mortality measurement is used in some of the quality indicators of the Agency for Healthcare Quality Research (AHRQ) and Premier Hospital Quality Incentive Demonstration. Maryland selected the APR-DRG grouper for severity adjustment of their Medicaid payment system and the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) have used APR-DRGs in the hospital accreditation process (Medicare Program, 2006).
The website www. healthgrades. com which provides comparative hospital and provider ratings uses APR-DRGs in addition to its own methodology in reporting quality data. Finally, US News & World Report magazine uses APR-DRGs to rank America’s best hospitals (Maurer, 2006). Are APR-DRGs effective as a Severity Adjustment Methodology? As indicated previously, the APR-DRG system is used by almost all the state data organizations and health agencies currently reporting comparative hospital data and have undergone the most extensive comparative and validity testing of any system.
While this fact alone does not definitively establish APR-DRGS as an effective severity adjustment methodology, it does strengthen the argument. A series of studies by Iezzoni and colleagues (1995, 1996, 1997) have compared the performance of APR-DRGs and other severity adjustment systems. They reported that APR-DRGs performed better than many competing systems in predicting inpatient mortality through severity of illness on a number of conditions such as coronary artery bypass graft surgery and stroke, but worse than other systems in looking at acute myocardial infarction and pneumonia.
Overall the findings suggest that no severity adjustment system, based on either administrative or clinical data, is clearly superior for all conditions and procedures. The optimal system for one condition may not be optimal for another. APR-DRGs perform about as well, and in some cases better, than competing severity adjustment systems based on administrative data. Given this variation in findings related to the studies of APR-DRGs and other systems, why are they so widely used?
APR-DRGs are widely used for severity adjustment today because of their low cost, flexibility, and methodology. Data collection for the APR-DRG system is based on standard, widely used abstract (administrative data) systems for individual hospital discharges (specifically, the UB-92). Reliance on currently collected data and information technology systems minimizes additional IT resources and data collection costs. In addition, the system has already been applied to a relatively extensive variety of discharge ata, allowing more complete construction of comparative provider and area norms. For many hospitals and states, the incremental data collection costs would be substantial for other related and well validated severity adjustment systems (Iezzoni et al. , 1995). APR-DRGs provide flexibility by permitting evaluation of both resource use (with a severity of illness classification) and outcomes (with a risk of mortality classification). In addition, the refinement incorporates a classification system for neonates.
Therefore, the system is theoretically appropriate for the distinctive disease characteristics of patients from all the groups. Finally, the general DRG methodology and medical terminology is widely understood by hospital administrators and physicians, which facilitates acceptance. The APR-DRG logic is open code, which permits the manual coding and checking of individual medical charts if necessary, and does not depend on specific computer software for implementation. 3M provides extensive user documentation and support, and has working relationships in place with most current and potential users.
Through APR-DRGs, hospitals, consumers, payers, and regulators can gain an understanding of the patients being treated, the costs incurred, and within reasonable limits, the outcomes expected. APR-DRGs also offer the opportunity for improvement in efficiency and identification of areas with potential quality problems. As the APR-DRG methodology continues to evolve, the role and importance of APR-DRGs in hospital quality and outcome comparisons will continue to expand. References Averill, R. F. , Goldfield, N. I. , Muldoon, J. , Steinbeck, B.
A. , & Grant, T. M. (2002). A closer look at all-patient refined drgs. Journal of AHIMA, 73 (1), 46-50. Boucher, A. , Bowman, S. , Piselli, C. , & Scichilone, R. (2006). The evolution of drgs. Journal of AHIMA, 77 (7), 68A-C. Bowman, S. (2006). New drg system for ipps: cms proposes severity-adjustment drg system based on apr-drgs. Journal of AHIMA, 77 (7), 18,20. Hall, D. , & Siegal, B. (2005). Taking the measure of measures: quality data initiatives and the challenge of effective and efficient data. Journal of AHIMA, 76 (2), 42-46.
Iezzoni, L. I. , et al. (1995). Predicting who dies depends on how severity is measured: implications for evaluating patient outcomes. Annals of Internal medicine, 123 (10), 763-770. Iezzoni, L. I. , et al. (1996). Judging hospitals by severity-adjusted mortality rates: the influence of the severity-adjustment method. American Journal of Public Health, 86 (10), 1379-1387. Iezzoni, L. I. , (1997). The risk of risk adjustment. Journal of the American Medical Association, 278 (19), 1600-1607. Maurer, W. G. , & Hebert, C. J. , (2006).
Cleveland Clinic Journal of Medicine, 73, S30-S35. Medicare Program: Changes to Hospital Inpatient Prospective Payment Systems and Fiscal Year 2007 Rates, 71 Fed. Reg. 47898 (2006) (to be codified at 42 C. F. R § 409, 410, 412, et al. ) McGlynn, E. A. , Damberg, C. L. , Kerr, E. A. , & Brook, R. H. , (1998). Health Information Systems Design Issues and Analytic Applications. Washington, D. C. : RAND. 3M Health Information Systems. (2003). All Patient Refined Diagnosis Related Groups (APR-DRGs); Methodology Overview. Wallingford, CT: Author.