PMPM expense data were obtained through patient-specific billing and utilization data. All of the subjects in our study were participants in a 100% capitated health care system. All office visits, pharmacy expenses, emergency department visits, and hospital admissions and procedures were tabulated. The expense data were categorized as total expenses, in-patient hospital care, pharmacy, and outpatient care. The $7 estimated to fund this program on an ongoing basis was added to the PMPM expenses of the experimental group.
Statistical Analysis
Differences between the experimental and control groups for fasting blood levels and PMPM expense data were assessed using a Student t test, both at entry and after 12 months. No statistical differences were noted between the control and experimental groups at entry. At entry and at 12 months we analyzed food intake questionnaires and laboratory results using an independent samples t test. Statistical analysis was performed using SAS software (SAS Institute, Inc, Cary, NC).
Results
Food Intake Questionnaires
Table 2 shows the mean fruit and vegetable intake for patients in the experimental group and those in the control group at entry and after 12 months. The experimental group patients increased their vegetable and fruit intake significantly compared with the control group over 12 months (P = .0001 and .0072, respectively).
The experimental group reduced their total fat intake and their saturated fat intake after 12 months; however, these differences in change were not significant (P=.4045, P=.1049).
Separate from the WHI food intake questionnaire, subjects were asked to report their weekly intake of legumes (a serving size was equal to 0.5 cup legumes) and the type of fat used for cooking and baking. The experimental group also reported a significant 45% increase in use of monounsaturated cooking oils compared with the control group’s 1% increase (P=.0001).
Fasting Blood Levels
Using a paired comparison t test, the experimental group noted a significant reduction in LDL levels, 117 mg per dL at entry to 104 mg per dL at 12 months (P=.0035,) while the control group’s LDL reduction was not significant, 119 mg per dL at entry and 111.7 mg per dL at 12 months (P=.1475). The difference in LDL reductions between the 2 groups was not significant by an independent samples t test. The total cholesterol/HDL ratio, Hb A1C, and triglyceride levels decreased for both groups; HDL increased for both groups, but the difference between the changes was not statistically significant.*
PMPM Expenses
Both groups noted reductions in PMPM total and in-patient expenses. The total PMPM expenses decreased 38% for the experimental group and 10% for the control group. No statistically significant differences were found between the groups’ total PMPM expenses (P=.2975).†
After the 1-year intervention, there was no difference in overall pharmacy PMPM expenses between the 2 groups (P=.4578). Specifically, lipid-lowering medication use and expense was very similar in the 2 groups before and after the intervention.
Discussion
In our small group of 97 patients with known CAD and elevated lipid levels, this intervention was not powered to yield significant improvements in clinical outcomes. Our study was associated with increased fruit and vegetable intake (nearly 2 more servings per day), a small increase in legume intake, and a switch to oxidation-resistant monounsaturated cooking oils.
Our study does help to confirm that half of patients with CAD are willing to make significant lifestyle changes when offered a program that emphasizes adding healthy foods in a group visit format. We also targeted patients with known CAD and elevated lipid levels and demonstrated that patients in the greatest need of therapy were willing to try lifestyle changes.
The experimental group noted significant reductions in LDL levels, but they were not statistically greater reductions than the control group. A bigger trend toward improvements in lipid and Hb A1C levels and reductions in total and saturated fat intake occurred in the experimental group than in the control group, yet these differences were not statistically significant. A 10% dropout rate was anticipated over the study; however, the additional 19% dropout rate that occurred after recruitment and before randomization was unexpected and limited the power to assess some of the trends we noted. The difference in LDL reductions between groups would have reached statistical significance if a limited reduction in LDL levels had occurred in the control group (<2%), or if the study had achieved the original target size planned and the changes noted persisted in the missing subjects.
The lipid reductions that occurred during our study may seem limited. However, this study reflects the ongoing implementation of various strategies over several years to improve lipid levels in this cohort that initially had LDL levels well above their target. In the 9-month time interval between identifying the patients and starting the intervention, LDL levels decreased by 16.5% in both groups, from a mean of 3.7 mmol per L (142 mg/dL) to 3.1 mmol per L (118.5 mg/dL).