MCOs, health plans may undercount your HEDIS performance data

Study reveals flaws in HEDIS comparisons of doctors

Your rate of mammograms, immunizations, and other preventive care measures provides a profile of "quality" that is reviewed by managed care organizations. But you may be doing a better job than they think.

The data collected by health plans as a part of Health Plan Employer Data Information Set (HEDIS) are not always statistically valid or accurate when viewed at the group practice level, according to a recent study in California that showed statistically significant disparities in three of six preventive care performance measures.

HEDIS data collected at the group practice level revealed higher rates for cervical cancer screening, mammography screening, and prenatal care than the aggregated data collected at the health plan level, according to the study, which was conducted by the California Cooperative Healthcare Reporting Initiative (CCHRI), a statewide collaborative of health plans, purchasers, and providers. Differences in rates were not statistically significant for childhood immunizations, cholesterol screening, and diabetic retinal exam. The study involved 10 medical groups and independent practice associations who collected HEDIS data during 1995.

Yet rather than spurring physicians to reject HEDIS altogether, the study should prompt physicians to collect and monitor their own HEDIS data and to try to influence the way they are calculated, says Mary Denton, RN, chief administrative officer for medical services at the Cedars-Sinai Medical Care Foundation in Beverly Hills, CA, and chairwoman of the CCHRI provider committee.

"Care happens at the provider level," says Denton. "Eventually, whether we like it or not, [consumers, employers, and health plans] are going to be looking for HEDIS data at the provider level. As providers we need to accept that and be a part of how that happens."

That sentiment is echoed by Jeff Rideout, MD, MA, vice president of quality management for Blue Cross of California in Walnut Creek. Despite its flaws, HEDIS represents an important first step toward standardized performance measurement designed to improve care, he says. He stresses that while plans want provider-level information, HEDIS isn’t being used to rate care at the group or physician level.

"HEDIS is a barometer of health plan performance," he says. "If it’s going to be used as a monitoring device for physicians, it’s seriously flawed. I have not seen one example of a medical group or physician being accountable for a HEDIS rate."

Why physicians dislike HEDIS

Physician discomfort with HEDIS begins with the measures themselves. They are "process" measures, monitoring what physicians or other health providers do, rather than clinical outcomes demonstrating the effect of treatment.

While physicians may prefer to have information on clinical outcomes, the items measured by HEDIS, such as mammograms and prenatal care, clearly are important preventive health issues, says Alison P. South, consulting manager of The MEDSTAT Group in San Francisco. MEDSTAT, a consulting and information firmed owned by Medical Economics, conducted the study and audits HEDIS data collection for the CCHRI collaborative. "It’s hard to argue with the legitimacy of some of the measures, even if they’re not true outcomes," she says.

HEDIS also requires physicians to take a public health approach to medicine, notes South. "Providers have been looking at one patient at a time," she says. "They need to [also] look at populations."

Data collection an ordeal?

Collecting HEDIS data has been a struggle for both health plans and group practices. Health plans typically extract data from claims, other administrative databases, or from a medical record review. Areas such as California that have a high degree of capitation require a greater reliance on medical record review. Instead of filing claims to health plans, provider groups report encounter data, a system that is far from foolproof, notes Denton.

For example, if a group practice subcontracts with a radiology center to perform mammograms under capitation, the clinical report may arrive and become a part of the patient chart, but the encounter report with the appropriate CPT code may not be sent to the group or the health plan, she says.

When health plans pull medical charts, they use a limited sample size, says David Hopkins, PhD, director of health information improvement for the Pacific Business Group on Health in San Francisco, a member of CCHRI.

"We were drawing samples of 400 [for HEDIS collection] at the health plan level," says Hopkins. "The largest provider might have 50 of those cases. Often, it’s 25."

One lesson from the CCHRI study, says Hopkins, is that group practices need to monitor their own HEDIS measures. "Providers should maintain complete and accurate data on their patients in an electronic form," he advises. "They should be aware of the items measured, and they should continually try to improve their performance on those measures."

Physicians also need to improve the data they report to health plans, even in such basic areas as complete and accurate coding, says Rideout. Both the quality of the data and of performance on HEDIS measures are highly variable among medical groups, he says. (For more information on how to improve your HEDIS data, see related story, above.)

"There is clearly a gap between what is happening and what is showing up on the data," he says. "If I don’t have the data, I can’t make conclusions about who’s doing a good job and who’s not."

Aside from detecting difficulties with data collection, the CCHRI study revealed some problems with HEDIS measures that are relevant nationwide:

Providers are held accountable for patients who recently enrolled with their group.

Health plans monitor HEDIS data for patients who have been continuously enrolled with the plan for at least one year. But if a patient spent 11 months with one group practice, then changed during the month of a HEDIS review, the patient’s failure to get a mammogram would be held against the current group, says Denton. (For more information on patient refusal to receive treatments that HEDIS data measures, see related story, p. 53.)

Pilot reported data on long-term patients

In the pilot study, group practices reported HEDIS data for patients who had been continuously enrolled with them for three years, which may account for some of the improved performance in the HEDIS measures. (See chart on p. 52.)

In the 10 medical groups and independent physician associations involved in the study, half of the women giving birth as patients of the practice were not continuously enrolled with that provider for the full year before their delivery. Many of the women had received their first trimester prenatal visit with another group, but the HEDIS data indicated a failure on this measure on the part of the current group.

HEDIS prenatal care data collected at the group practice level was 20% higher than the data collected by health plans.

At Blue Cross of California, Rideout says he applied the continuous enrollment rule to medical groups and IPAs before analyzing their aggregated HEDIS data. Still, he says he found wide variations in HEDIS performance, from very high rates of care to near zero.

The target population included patients for whom the preventive care was inappropriate.

When a patient was evaluated as a possible diabetic, the physicians often used a diabetic code. Yet even if the tests turned out negative, these patients then became a part of the patient population expected to receive a diabetic retinal exam.

The CCHRI study showed that an average of 35% of the diabetic retinal sample did not have insulin-dependent diabetes, which is the appropriate population for this preventive exam.

Women who had hysterectomies were included in sample populations for Pap smears, and children who had illnesses that were contraindications for immunizations were included in that baseline population.

Some concerns about the HEDIS populations have been addressed by the newly created Practicing Physician Advisory Council of the National Committee on Quality Assurance (NCQA) in Washington, DC, the health plan accreditation organization that developed HEDIS. "There was considerable physician input into the methodological efforts [of HEDIS 3.0, which was released last year]," says Bernard M. Rosof, MD, FACP, senior vice president of medical affairs at Huntington (NY) Hospital and a member of the NCQA Practicing Physicians Advisory Council.

Rosof, who is president-elect of the American Society of Internal Medicine in Washington, DC, says NCQA is making "an honest effort to obtain the physician community’s advice and to have their buy-in."

Accurate data collection is difficult, especially when care occurs outside the group practice.

Public health clinics frequently offer free immunizations. But for those to be counted toward the group practice’s HEDIS measurement, the group needs specific information from the clinic, such as the name and professional designation of the physician or nurse who gave the injection and the lot number of the vaccine.

To collect accurate HEDIS data in other areas, primary care physicians need the help of radiology centers, labs, and OB/GYNs.

These and other problems with HEDIS collection should inspire group practices to examine their information systems, improve communication and data exchange with outside providers, and seek solutions to HEDIS quandaries, says Denton. In the next phase of CCHRI’s work, practices throughout California have been asked to collect their own data on diabetic retinal exams and hemoglobin A1(c). Eventually, providers may have the option of reporting their own, audited HEDIS data, which would be more accurate than the health plan data, Denton says.

Practices are wasting their energy if they simply try to fight the use of HEDIS data. Instead, says Denton, they should "figure out how to win, how to work with administrative data, how to work with physicians, how to educate patients, and how to be more proactive."

Rosof suggests physicians begin by designing an internal quality improvement program: choose benchmarks of care (such as HEDIS), measure them, assess performance, identify and implement changes in practice, then reassess.

Physicians need to take a businesslike approach to their practice by investing in information systems for data collection, advises Jonathan Seltzer, MD, director of health systems research for Premier Research Worldwide in Philadelphia and author of Models for Measuring Quality in Managed Care: Analysis and Impact (Faulkner & Gray; 1997).

If HEDIS data is reviewed by health plans, or even possibly by lawyers in a malpractice suit, group practices should be one step ahead with monitoring and quality improvement practices, he says.

"There’s going to be a cost of gathering data," says Seltzer. "But what’s the cost of getting kicked out of your HMO? Or of a lawsuit? There’s a change in the way medicine is practiced, and you’re accountable now."

[Editor’s note: For a copy of the Provider Pilot Report, contact The MEDSTAT Group, c/o CCHRI, One Sansome Street, 15th Floor, San Francisco, CA 94104. Telephone: (415) 732-6240. Fax: (415) 732-6204.]