Clinical trials provide strong support for using 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitor (statin) therapy in patients with coronary heart disease (CHD) (1-7). Statins are a relatively safe and well-tolerated class of agents that have proven to be effective in treating lipid disorders. Recent data suggest, however, that statin therapy remains underused in the treatment of lipid disorders, or dyslipidemia. Reasons for this underuse of statins include inadequate physician titration of medications to reach the goal level of low-density lipoprotein cholesterol (LDL-C) recommended by the National Cholesterol Education Program (NCEP) and inadequate long-term patient adherence to prescribed drug therapy (8-10).

In this study, we assessed trends in the use of statin therapy and changes in LDL-C levels in a well-defined population of adults with CHD receiving their care at a single large multispecialty medical group. During the 4-year study period, the medical group emphasized the importance of lipid control in CHD patients. Primary care physicians had unrestricted access to several statins through the medical group's drug formulary. Clinical guidelines for lipid control emphasized aggressive pharmacotherapy, and patients received messages on the importance of lipid control through periodic medical group publications sent to their homes. Our study objective was to determine whether systems of care implemented within a large medical group are associated with improved treatment and control of dyslipidemia in a high-risk group of CHD patients.

The overall population of individuals with at least two qualifying LDL-C measures (n = 1388) is described in Table 1. Although baseline and follow-up lipid measures were required to be at least 365 days apart, the median time between the measures was 917 days, or 2.5 years.

The mean age of study subjects with two qualifying LDL-C measures was 64.9 years; 66.6% were male, and 24.1% had a college degree. Current smoking was reported by 7.1%, and 59.4% indicated a prior history of smoking. Regular aspirin use was reported by 82.9%. Mean body mass index (BMI) was 27.1 kg/m², and 24.3% were identified as having diabetes at baseline. There were statistically significant differences between study-eligible members with and without two qualifying LDL-C measurements for all variables other than BMI and baseline diabetes.

Table 2 describes LDL-C and HDL-C measures at baseline and at follow-up. Data are presented for all patients with qualifying LDL-C measurements (n = 1388), for patients with baseline LDL-C >100 mg/dL (n = 1214), and for patients with baseline LDL-C >130 mg/dL (n = 781). For all patients with qualifying measurements, there was a significant decline in mean LDL-C, from 137.6 mg/dL at baseline to 111.0 mg/dL at follow-up, a change of -26.6 mg/dL (P < .001). Mean HDL-C also improved, with an increase from 42.3 mg/dL at baseline to 46.3 mg/dL at follow-up, a change of +4.0 mg/dL (P < .001). For the subgroup with baseline LDL-C >100 mg/dL, mean LDL-C decreased from 144.7 mg/dL at baseline to 113.5 mg/dL at follow-up, a change of -31.2 mg/dL (P < .001). Mean HDL-C in this subgroup increased from 42.8 mg/dL to 46.6 mg/dL, a change of +3.8 mg/dL (P = .008). For the subgroup with baseline LDL-C >130 mg/dL, the change in mean LDL-C was -42.5 mg/dL (P < .001), and the change in HDL-C was +3.5 mg/dL (P = .02). The percentage of patients in the study with LDL-C ≤100 mg/dL rose from 12.5% to 39.8% (P < .001), and the percentage of patients with HDL-C ≥40 mg/dL rose from 52.5% to 67.6% (P < .001).

Table 3 shows the proportion of patients receiving statins (among the 1038 subjects with pharmacy coverage and two qualifying LDL-C measures) at both baseline and follow-up and the subsequent discontinuation rate more than 1 year later in two groups of statin users: patients who used statins at baseline and patients who used statins at any time during the study period. Statin use among all study subjects in this group was 24.3% at baseline and 69.6% at follow-up. In those with baseline LDL-C >100 mg/dL, statin use was 21.0% at baseline and 70.4% at follow-up. In those with baseline LDL-C >130 mg/dL, statin use was 16.7% at baseline and 74.2% at follow-up. Statin discontinuation rates were 8.3% for baseline statin users, 9.5% for baseline users whose baseline LDL-C was >100 mg/dL, and 4.1% for baseline statin users whose baseline LDL-C was >130 mg/dL. For subjects who used statins at any time during the study period, the discontinuation rate was 12.2%.

Table 4 reports the multivariate analysis of LDL-C and HDL-C changes and statin use for the 1038 study-eligible members with pharmacy coverage. The model presented in Table 4 includes administrative data only. The model presented in Table 5 adds survey data and is limited to the 804 individuals included in the previous model who returned surveys with completed responses on variables of interest. Residual plots showed neither excessive nonnormality in the residuals nor heteroskedasticity.

In the models of Table 4, baseline LDL-C, sex, and statin use are significant predictors of decreases in LDL-C. Table 5 shows that use of statins for the entire period between baseline and follow-up measures (i.e., statin use = 1.0) was associated with an additional 17 mg/dL decrease in LDL-C compared with no statin use. On average, men experience an additional 7 mg/dL decrease in LDL-C compared with women. Only baseline HDL-C and sex are significantly related to changes in HDL-C. On average, women experienced a 1.7 mg/dL greater increase in HDL than did men. Among the variables collected from surveys (Table 5), aspirin use was associated with a statistically significant (6 mg/dL) decrease in LDL-C, and a 1-point decrease in BMI was associated with a 0.29 mg/dL improvement in HDL-C.

We analyzed trends in LDL-C and HDL-C levels in patients with diagnosed heart disease having two LDL-C measurements in a large medical group. Statin use was analyzed for the subgroup of 1038 (75%) with pharmacy coverage. At baseline, all patients had established CHD and hence would have had an NCEP-established goal of LDL-C <100 mg/dL at the time of the study. However, at baseline only 24.3% were receiving statin therapy, and only 12.5% had LDL-C of ≤100 mg/dL. After a mean follow-up period of 2.5 years, mean LDL-C and HDL-C levels had significantly improved in the entire cohort, as well as in subgroups stratified by baseline LDL-C of >130 mg/dL and >100mg/dL. There was a corresponding marked increase in statin use during the follow-up period, with a significant increase in patients achieving the study goal of LDL-C ≤100 mg/dL (12.5% at baseline vs 39.8% at follow-up, P < .001).

These data document a significant improvement in the proportion of CHD patients treated with statins and reaching their NCEP-established LDL-C goal in the late 1990s (19-21). Our results favorably compare with contemporaneous data collected on the general Minnesota population, which showed worsening cholesterol levels in the 1990s (8). Results also favorably compare with the survey of 48,586 individuals with coronary artery disease from practices throughout the United States, chosen for their frequent use of cholesterol-active medications, in which 44% had annual testing of LDL-C. Of those who had annual testing, only 25% reached the target LDL-C of ≤100 mg/dL, and only 39% were taking lipid lowering therapy (9).

The Lipid Treatment Assessment Project (L-TAP) was a comprehensive survey of lipid therapy prescribing habits and lipid results conducted in five regions of the United States and drawn from a group of primary care physicians, also targeted because they wrote large numbers of prescriptions for lipid-active drugs (10). The L-TAP study demonstrated that in a cohort of 1460 patients with CHD, only 18% attained the NCEP-established goal of LDL-C <100 mg/dL, compared with 39.8% in our study who attained an LDL-C level of ≤100 mg/dL. This was despite the fact that 84.6% received treatment with lipid lowering agents, compared with 69.6% receiving statin therapy in our study (10).

While baseline LDL-C levels were being measured for this study (median date of October 12, 1995), a system-wide program of health-related goals was implemented at HPMG. Among these goals was a heart-health goal that called for a 25% reduction in CHD events within 4 years (22). The importance of lipid control in CHD patients was emphasized by medical group leaders in meetings with primary care physicians and certain subspecialty physicians, including cardiologists.

During the years of this study, systems of care were developed and deployed within the medical group to support both patient and provider attention to control of lipid disorders (22-26). Results of lipid testing were stored for electronic retrieval via computer. Lipid test results, with an explanation and recommendations, were mailed to patients. High-risk patient registries that included adults with CHD or diabetes and LDL-C test results or indicated a need to obtain an LDL-C test were provided to clinics and physicians as an aid for tracking, visit planning, and active outreach. One of the authors (RJG) gave a series of lectures at each medical group clinic about the use and benefits of lipid-lowering therapy. Also, a specialized program known as Lifestyle Management was added to the cardiac rehabilitation program. This program featured one-on-one sessions with nursing staff for patients and follow-up case management, as well as a direct review of each patient's care plan and progress by the medical director. During the study period, it also became commonplace to initiate statin therapy before hospital discharge following an acute cardiac event, such as a myocardial infarction. These programs, initiated to aggressively reduce cardiovascular events in high-risk patients during the study period, likely accounted for some of the observed improvement in lipid control. It is unlikely that secular trends accounted for all the improvement, because few other medical groups have achieved similar levels of lipid drug use (27) or lipid control (28), and lipid trends in Minnesota during these years were not improving (8).

Several factors limit the interpretation of the data presented here. First, misclassification of CHD status is a possibility; however, we used a validated method for heart disease identification with an estimated sensitivity of 0.85 and a positive predictive value of 0.89 (13). Second, the study was limited to one large medical group in Minnesota. Studies in a variety of other settings may be needed to replicate and extend the findings reported here. Finally, the requirement that each patient have at least two LDL-C measurements over an average 30-month period may have identified a population of patients having more active management of dyslipidemia. However, this selection strategy was essential, given our intent to assess trends of LDL-C and statin use over time. We were unable to compare patients with and without qualifying LDL-C measures in relation to baseline LDL-C levels or statin use. We observed at least one LDL-C in only a portion of those without two LDL-C measurements, and the majority of the single LDL-C measures that exist were observed near the end of the 4-year study period. Therefore, the available LDL-C measures for the group excluded from the analysis do not provide an analogous baseline LDL-C measure. Likewise, our measure of statin use is tied to LDL-C baseline and follow-up measures and could not be computed for participants without two qualifying measures.

We conclude that substantial improvement in LDL-C and HDL-C control occurred in adults with CHD at this practice site during the late 1990s. The main cause of the improvement was a dramatic increase in statin use, which was significantly related to LDL-C change. However, other factors contributed to improved lipid control in this population, including coordination of use of clinical guidelines among clinics, use of high-risk patient registries, and use of automated monitoring and prioritizing of patients for special attention. As LDL-C goals become more stringent (3,4,6), the application of effective outpatient chronic disease care strategies such as registries, active outreach, visit planning, and coordination of care across sites will increase in importance. These data show that primary care clinics and providers are capable of dramatic improvements in care over a short period and suggest that resources invested to improve outpatient care can rapidly return a sizeable clinical return on investment.