Johns Hopkins ' ACG Risk Adjustment System Utilizes Predictive Modeling

Health plans and insurers increasingly use sophisticated risk-adjustment tools to manage quality care and control health costs. One of these tools is the DST Health Solutions-distributed Johns Hopkins Adjusted Clinical Groups Case-Mix System (the ACG System), the most widely used population-based diagnostic risk-adjustment system in the world. More than 250 organizations in the global healthcare marketplace have adopted the ACG System to help them manage care for tens of millions of patients.

The ACG System is a risk measurement methodology that allows managed care organizations to describe a population's healthcare needs using either diagnoses, prescriptions or both. In a climate where managing costs and improving efficiency and patient care are critical initiatives, the ACG System enables health plans to balance both objectives. Using the system's unique suite of tools, health plans are able to obtain key information about their member populations to help them negotiate more equitable contracts with physicians and employers. The system creates a common language of healthcare analysis that benefits providers, purchasers and consumers of healthcare.

DST Health Solutions and the ACG Development Team at the Johns Hopkins Bloomberg School of Public Health have incorporated additional predictive modeling capabilities into the latest version of the ACG System. This tool will helps health plans and administrators quickly identify members who are likely to be high risk or resource intensive in the future, thereby improving care while reducing potential expenses.

The science of the ACG System

In 30 years of experience developing state-of-the-art case-mix and risk measurement tools, the research team at the Johns Hopkins Bloomberg School of Public Health found that very few people have just one health condition over an extended time period. The nature, scope and potential interaction of multiple conditions are, however, key driving factors in assessing a person's risk for high healthcare resource use. This finding led to a new method of classifying and categorizing patients for analysis where patients' total incidence of disease, or morbidity, is used to predict health service resource use instead of evaluating the diseases as separate events. Classifying a patient using this method better reflects how patients present to the healthcare system and provides a more accurate depiction of patient health status.

Since its original development in 1991, the ACG System has been validated extensively and refined based on feedback from hundreds of health plans covering hundreds of millions of lives. More than 100 academic groups around the world have tested ACGs, and numerous peer-reviewed articles validating or applying the system have been published. (Read a case study on CareOregon .)

The ACG System uses a comprehensive categorization approach based on the premise that the level of resources necessary for delivering appropriate healthcare to a population is associated with the morbidity burden of that population. The system relies on the diagnostic code information found in professional and hospital insurance claims or other computerized records and the National Drug Codes (NDC) present on retail prescription claims. This provides the user with a more accurate representation of how people present to the healthcare system-as a constellation of morbidities, not as individual diseases. Through ACGs, patients are classified into groups based on commonly occurring patterns of morbidity that span multiple disease categories-thus providing a far more accurate picture of the complex interplay of the elements of a patient's health.

What separates the ACG System from others is its ability to reflect the multifaceted nature of coexisting conditions experienced by the majority of persons enrolled in a health plan. Alternative methodologies, which treat individuals as merely the sum of their diseases, fall short because they do not adequately account for variability of cost across individuals with multiple co-occurring morbidities.

Dr. Jonathan Weiner, ACG Team Leader and Professor at the Johns Hopkins Bloomberg School of Public Health, explains that counting individual diseases is not the most effective way to measure overall clinical need, either future or present. "Knowing that a person has a specific [usually major] disease or condition can be important for managing their care and predicting future resource use," says Weiner. "Knowing which patients have acute conditions that will resolve with treatment and which patients have chronic conditions leading to declining health puts the provider or health plan in a more powerful, proactive position to manage the patient more effectively-and to gauge the expense."

Health plans can use the same analysis to assess the performance of healthcare providers in the network, a form of physician profiling. By accounting for differences in patient case-mix, the plan can sort out factors caused by practice patterns from factors caused by underlying morbidity. With this information, health plans can create incentive programs that encourage providers to manage patient risk effectively while identifying potential medical and pharmaceutical fraud and abuse. Additionally, providers with a sicker panel of patients are not penalized for the resulting increase in resources. With the use of the ACG System, the reimbursement process can be made more equitable and health plans can experience higher retention of quality providers within their networks.

Employers, too, benefit from the data produced by the system. Plans may intuit that a blue-collar employer will likely have a high incidence of accident claims while a white-collar company is more apt to report carpal tunnel, but the ACG System can provide the employer with data and justification. "Being able to pinpoint scientifically where the cost will be coming from and having a method to keep employees healthy helps employers manage their own premium costs and the premiums, deductibles and out-of-pocket expenses passed on to their employees," says Steve Sabino, vice president of DST Health Solutions' Health Plans Solutions Group.

In addition to working with prominent commercial insurers and health plans, integrators of decision support systems, healthcare consultants and researchers, the ACG System is also used to exchange hundreds of millions of dollars on behalf of millions of covered enrollees in both private and public sector managed care programs. Public programs in the U.S. include Alabama, Arkansas, Kansas, Maryland, Minnesota and Oklahoma state Medicaid programs. ACGs have also been used by the U.S. Department of Veterans Affairs and have been adopted for province-wide physician profiling in British Columbia, Canada .

Predictive modeling for a changing healthcare environment

"It used to be that physicians could charge a health insurer for virtually any service delivered to a patient and it would be paid," explains Sabino. To help control the use of expensive specialty and acute care, health maintenance organizations (HMOs) began to use primary care physicians as "gatekeepers." Today, with the recent consumer backlash against managed care, the decline of the gatekeeper model and the movement toward consumer choice in healthcare, health plans are looking for new approaches to control medical costs. "We're seeing a greater need among health plans and employers for information that helps them to be more proactive about keeping members healthy and productive and thus avoid the unnecessary use of costly medical services and procedures," says Sabino.

In response to the healthcare industry's need for tools to help them be proactive with regard to patients' health, DST Health Solutions and the ACG Team have focused on predictive modeling. Predictive modeling is an automated risk assessment tool that brings efficiency to case selection and case preparation for care management as well as efficiency and improved accuracy to small group underwriting renewals.

Predictive modeling allows healthcare organizations to target patients who would benefit from case management, a personalized, interactive process to manage disease preventively before it results in costly care. With the cost of healthcare rising each year, predictive modeling can help align premium levels with the risk of the employer group. Because the ACG System can stratify members within a disease category, health plans can adjust care and resources to match the degree of care needed. If, for instance, a health plan has a concentration of women over a certain age with diabetes, the ACG System stratifies the women by risk, allowing the health plan to assess higher-risk women. Once identified, the plan may direct healthcare personnel and administrators to proactively monitor diet and other indicators that can prevent major complications, a version of case management.

For more about DST Health Solutions, contact us at inforequests@dsthealthsolutions.com.