Raising the health status of HMO members: several HMOs are using a statistical model to measure access, morbidity, and utilization - includes analysis of how Blue Cross and Blue Shield of Nebraska, Samaritan Health Plan and Harvard Community Health Plan are approaching the concept

Business & Health, April, 1994 by Manon Spitzer, John Billings, Kathryn Coltin, Arlene Davidson, Sally Hrdy

Several HMOs are using a statistical model to measure access, morbidity, and utilization.

Employers often feel competent to judge a managed care system's performance in meeting cost containment objectives based on premiums. Employers feel less competent, however, at evaluating the same system's success in improving its participants' health.

Recently, insurers and employers have begun to use statistical monitoring techniques to measure the effect of a health care system on a defined population and to assess how specific providers perform.

Benefit and health plan managers can use such information to introduce "prospective" or "anticipatory" case management. This strategy aims to enhance health system quality and effectiveness while containing overall costs.

Of the three HMOs discussed in this article, one learned that preventable hospitalizations from some subscribers cost nearly 50% more per employee than for others with the same conditions. The second HMO discussed learned that its aggressive utilization management program resulted in admission rates for medical conditions that were 20% below the age- and sex-adjusted average of the general population. And the third HMO learned that its low admission rates for certain conditions were not caused by a biased selection that favored healthier individuals.

The statistical monitoring approach developed by John Billings, of New York University, and by Codman Research Group Inc., a company specializing in health care information systems, complements and expands the scope and validity of the Health Plan/Employer Data and Information Set (HEDIS 2.0) performance standards in two ways. (HEDIS, popularly known as a report card, proposes standardized measures to help employers evaluate health plans.)

First, statistical monitoring takes into account differences in certain factors--principally morbidity--that affect utilization and costs. This type of sensitivity makes it possible to compare health plans for city employees, for example, with those covering university professional staff. Second, it produces information to explain why utilization differs from what was expected and how problems can be addressed.

The statistical monitoring system is embedded in CRG's software used by Blue Cross and Blue Shield of Nebraska (BCBSN) in Omaha; by the Samaritan Health Plan (SHP), an HMO in Phoenix,; and in CRG's processing of data for the Harvard Community Health Plan (HCHP), a 532,000-member HMO in Brookline, Mass. This approach also is being tailored for several employers.

How It Works

Most strategies used to evaluate quality assurance seek the "bad actor" by identifying what is "not quality," largely through case-by-case utilization review. Review activities target specific diseases and surgical procedures, largely high-cost, high-risk procedures performed in hospitals, and the performance of individual providers. Such data usually have little value for management because they seldom identify behavioral patterns that are consistent, statistically significant, and consequential. Most important, the information often results in actions that are more punitive than constructive.

Statistical monitoring differs from these traditional strategies in four important ways: 1) It seeks to generate information that supports efforts for improvement by providing feedback of clinically relevant information on behavior that is systematic and therefore subject to management; 2) It seeks to minimize intrusive micromanagement of providers or employees by identifying, then focusing on for interventions, only the "drivers" of variation; 3) It is conducted using established epidemiological methods, adjusting for demographic differences--such as age, sex, and morbidity--and applying tests of statistical significance to ensure that utilization and cost differences are not merely random fluctuations; and 4) By organizing systems to provide appropriate care at appropriate times, it achieves lower overall costs.

A central premise of the systems approach to quality evaluation is that rates of hospitalization for specific medical conditions and surgical procedures, age- and sex-adjusted to a practice norm, gauge the effectiveness of the health care delivery system. They are also indirect measures of the health of the members it serves. Data on only hospitalization rates are not enough to resolve every problem, but experience has shown that they raise the right questions so that detailed investigations, including review of ambulatory service use and provider practice patterns, can pinpoint the causes of a problem and focus on corrective actions.

Clinically specific hospitalization admission categories are divided into three groups for monitoring and evaluation:

1. Ambulatory care sensitive (ACS) medical conditions. This category is key to assessing the effective use of ambulatory care, principally primary care physician services, as well as general health status.

2. Referral sensitive surgical procedures. This category is key to assessing access to specialists and the willingness of cost-conscious health managers to provide expensive care when it is required. But it is often difficult to interpret; when rates are low and access is adequate, the data may reflect discretion in physician practice patterns or the good health of those who have less need for interventions.