For over two decades, the ACG® System has been the standard for risk adjustment. Based on the premise that clustering of morbidity is a better predictor of health services resource use than the presence of specific diseases or disease hierarchies, the ACG System provides a multi-morbidity framework that is clinically logical, informative of future healthcare resources, easy to use and applicable to both financial and clinical managers.
Some of the unique advantages of the Johns Hopkins ACG® System are:
Clinical complexity represents more than merely summing diagnosis codes. The fundamental difference between the ACG System and other case-mix systems is that the ACG System categorizes people. At times, many clinicians and health administrators think of patients as fitting into single diagnostic classes – asthmatics, diabetics, patients with heart failure. But measuring the health of a population using indicators of specific diseases or episodes-of-care fails to capture the entirety of morbidity and health care that a person experiences over time and across service settings. Information on the totality of a patient’s healthcare needs is required to anticipate resource consumption adequately.
Supports the full-spectrum of management applications. The ACG System describes the health status of all persons in a population, ranging from those with relatively minor medical needs, such as preventive services and acute infections, to the most expensive, sickest 5% of the population that consumes 50% of all health care resources. This characteristic of the ACG System makes it well-suited for several applications, including identifying high risk patient with multiple morbidities, describing the morbidity burden of a population, assessing primary care provider performance, profiling entire health systems, targeting groups of patients who would benefit from care management, and rate-setting.
Avoids basing complexity on specific procedures or hospitalizations. Unlike other case-mix/risk adjustment systems, ACGs do not use admissions or procedures to categorize people. Categorization is based on patient needs, as reflected in clinical diagnoses rather than in a provider’s practice pattern. Thus, the ACG System is not prone to the potential shortcomings of other systems that can categorize patients into “sicker” categories because a clinician’s practice style is more intensive. Moreover, the ACG System does not provide perverse incentives for clinicians to perform unnecessary procedures or hospitalize patients to increase their risk rating and thereby secure higher levels of compensation.
Provides more than a simple score. Throughout the many years that the ACG System has been available, repeated studies of the ACG System methodology have confirmed its predictive power relative to total cost, but studies also correlate ACGs with functional status, provider productivity, hospitalization and even death. The ACG System provides more than a simple statistically derived score; the ACG System is a complete taxonomy to describe population health. The ACG System also permits ready customization to reflect local data. This degree of transparency and localization is especially important in assessing provider performance as peer group comparisons are often the most effective comparisons to engage providers. The holistic population view of ACGs coupled with the ability to aggregate ACGs into relatively few morbidity categories, allows the ACG System to provide robust and statistically valid profiles of typical managed care provider panels.
Stable over time. Unlike predictive models based on neural networks or artificial intelligence techniques which fit a model to specific set of data, ACGs are based upon a well-defined set of clinical indicators that have been vetted in the health services research literature. This approach is not only more transparent and defensible than neural network models, it is also more stable as it is not dependent on the idiosyncrasies of a particular set of data.
Maximizes the use of available data. The ACG System provides models based on medical claims only, pharmacy claims only or combined medical and pharmacy claims. The analyst can choose the most appropriate model from a suite of predictive models based upon the available data, product line and intended purpose. Predictions of both total cost and pharmacy cost are available. We also offer models that predict hospitalization.
Flexible and Customizable. The ACG System is the only population-based system, which is totally adaptable to a local context. While our beginnings come from the US health care system, we have since developed the ability to take into consideration local cost structures, coding systems, practice behavior, and language adaptations as well as the availability of local markers whether they capture socio-economic, functionality, living arrangement, or other dimensions. These new variables can be placed within user-created Risk Assessment Variable Sets (RAVS) that can substituted for the supplied default case-mix weights and predictive models. The ACG System supports a wide range of diagnostic and pharmacy code sets, including ICD9-CM, ICD10-CM, NDCs, ICD10, ATCs, ICPC, and Read Codes.
