SECONDARY PREVENTION OF CORONARY ARTERY DISEASE

Targeting Lipid Management With Strong Interventions

Although prevention of disease can sometimes represent a significant expenditure of resources, a lack of prevention can prove to be even more expensive. Recognizing this cost and, most importantly, the health of its members, this health plan made a significant investment in lowering lipid levels for its patients with known coronary artery disease (CAD).

Although cardiac disease affected only 5.2% of this plan's population it accounted for $150 million of the plan's annual budget. At this plan, half of all myocardial infarctions (MI) occurred in members with a previous MI.

The Scandinavian Simvastatin Survival Study (4S) showed that treatment of patients with confirmed coronary artery disease using lipid-lowering agents could result in a decreased risk for coronary events.

An analysis performed by the plan showed that aggressive lipid lowering was not occurring in patients with confirmed CAD. Based on its analysis, the plan believed it could improve quality by reducing the risk of subsequent coronary events suffered by this population.

The targeted population consisted of enrollees with high lipid levels and a history of a previous CAD event in the last year. Claims and encounter data were used to identify this population. Members were included if they had (in the last 12 months) experienced:

• hospitalization for a documented myocardial infarction, or
• coronary angioplasty, or
• coronary artery bypass graft surgery

(Patients whose lipid levels were already optimal without lipid-lowering therapy were eliminated from the target group.)

The plan selected three performance measures for this activity:

• the percentage of patients who had a low-density lipoprotein (LDL) measurement performed within one year of a CAD event
• the percentage of patients whose LDL cholesterol was <130 mg/dL within 18 months of a CAD event
• the percentage of patients with a total cholesterol/high-density lipoprotein (HDL) ratio of <6 within 18 months after a CAD event

The plan used automated laboratory data for those patients seen in the plan's health centers. For the patients cared for by IPA physicians, these measurements were tracked by an annual chart review. No samples were used; all measurements were of the entire targeted population.

The baseline performance was measured in 1994. The plan found that 52% of the target population had received an LDL test, but that only 32% had an LDL level <130. Of the target population, 59% had a total cholesterol/HDL ratio <6.

The plan could find no benchmark data for these measurements.

The objective of this initiative was to increase the percentage of high-risk patients with high lipid levels who were aggressively managed and to lower their lipids to more acceptable levels. The plan established performance goals for two of the three measurements. The goal for the percentage of patients who had received an LDL test was set at 75%, and the goal for the percentage of patients with an LDL level <130 mg/dL was set at 52%. No goal was set for the percentage of patients with a total cholesterol/HDL ratio <6.

The plan performed a comprehensive barrier analysis using a fishbone diagram. This analysis identified a number of barriers that prevented the aggressive lowering of lipids:

• Patients visited multiple providers. With practitioners in family practice, internal medicine, and cardiology all caring for the same patient, it was unclear who was responsible for managing the patient's lipids
• There was no system to deliver information to primary care physicians (PCPs) about the population of these patients in their panels
• Providers were reluctant to prescribe high-cost medications without clearly identified guidelines
• The group's current formulary lipid-lowering agent was not capable of lowering lipids to the extent described in the 4S trial

After measuring its baseline performance, the plan was impressed with the large degree of variation in the performances from center to center and office to office. This appeared to reflect the varying degrees of emphasis each site gave to preventive services.

A cardiac risk reduction team was formed at each medical center. This team was composed of clinical and nonclinical staff and was held accountable for results at its center.

PHYSICIAN THOUGHT LEADERS Though the plan initially selected department chiefs to serve on these teams, it soon realized that the individuals in this position were not necessarily best suited for the tasks that the team was addressing. In the second year of the program, the plan revised its strategy and asked clinicians at each site to select their own champion to participate on the cardiac risk reduction team at each center.

The plan began its intervention by implementing guidelines, consisting of recommended visit schedules, laboratory tests, and ideal target results. It trained clinic staff on these guidelines via brown bag lunches.

The pharmacy and therapeutics committee placed a more powerful lipid-lowering drug on the plan formulary.

Next, the plan set up a patient registry and designed patient tracking forms. Originally a paper-based system, the patient registry contained the most recent laboratory and pharmacy data. It also contained demographic information on all the patients in the targeted population. This registry gave the PCPs their first look at their entire secondary prevention population. The plan also designed patient action forms. These forms were medical record sheets that listed the guidelines and interventions of the secondary prevention program. The forms enabled the tracking process by neatly organizing information related to lipid management in one convenient location.

SPECIFICITY, SENSITIVITY, CREDIBILITY The plan learned the importance of identifying the right patients. If a member was erroneously identified as being part of the target population, doctors quickly lost confidence in the data. The plan found that it was better to miss a few patients who should have been included rather than to accidentally include inappropriate ones. To maintain physician support, the data on every patient must be accurate.

Remeasurement, using 1995 data, showed that the percentage of patients who had received an LDL test had risen to 62%. This represented a statistically significant improvement. The percentage of patients with an LDL <130 had increased to 41%, also a statistically significant improvement. There was no change in the percentage of patients with a total cholesterol/HDL ration <6.

The plan was encouraged by this progress but wanted to get closer to its goals. It noted that some clinics had made marked improvements, while others continued to perform poorly. It identified a number of continuing barriers:

• the lack of automation of the registry
• patients who had coronary events prior to joining the plan had not been identified
• accountability was still unclear at the poorer-performing clinics

To address the issue of accountability, the plan continued its medical training via brown bag lunches, focusing on the poorer-performing clinics. It held several all-day training sessions for clinic staff.

It continued to give each clinic feedback on its performance. It began to report individual practitioner rates for the percentage of their CAD patients who had received an LDL test. Interventions for poor performers included counseling.

Remeasurement, using 1996 data, showed that the percentage of patients receiving LDL testing had increased to 70%. The percentage of patients with an LDL <130 had risen to 51%, and the percentage of patients with a total cholesterol/HDL ratio <6 had increased to 67%. All these improvements were statistically significant.

The plan maintained the interventions that it felt had been most effective to this point. These included the educational efforts focused on poorer performers and the all-day training of clinic staff. At the same time, the plan introduced a series of new interventions:

• It automated its patient registry. This allowed patients not in compliance to be flagged
• It conducted one-on-one coaching with staff not using the registry
• It implemented several half-day training classes. These classes reinforced the appropriate use of the guidelines and the registry
• It continued to improve the registry based on user feedback. Key improvements included the development of a patient summary report, which was printed when members in the registry came into the clinic. This tool reminded the providers of issues related to the member's heart disease. It served as a structured chart note, and encouraged the collection of data
• Physicians continued to receive individual feedback on the percentage of their patients who received an LDL test and the percentage of their patients with an LDL <130 mg/dL
• Chart stickers were developed to flag patients as registry members
• 1,700 additional patients were added to the registry (most had CAD events outside the plan)

In both the measures for which it set goals - the percentage of the population who had received an LDL test and the percentage whose LDL was <130 - the plan exceeded it goals. The final measurement, using 1997 data, showed a rate of LDL testing of 76.7%. Of tested patients, 63.4% had an LDL <130. Both these measurements represented statistically significant increases. (Plan staff questioned the value of the measure of total cholesterol/HDL ratio, and therefore the measure was dropped.)

The plan continues to focus on primary and secondary prevention. It has since implemented an incentive plan for all physicians, based on its rate of performance of ten preventive care measures, including the remaining two from this activity.

Based on the success of this activity, the plan adopted it as a model for improving care for another disease area - its diabetic population.

This profile demonstrates how to use barrier analysis effectively to design escalating interventions: covering all involved parties, repeatedly analyzing barriers and interventions, and continually searching for ways to further improve.