Efforts to Improve Subsequent Treatment of Cardiovascular Risk Factors in Older Patients With Diabetes Hospitalized for a Cardiac Event

December 1, 2005
Diana Echeverry, MD
Diana Echeverry, MD

,
Margie Dike, PhD
Margie Dike, PhD

,
Lois Jovanovic, MD
Lois Jovanovic, MD

,
Alison Okada Wollitzer, PhD
Alison Okada Wollitzer, PhD

,
Sydney Westphal, MD
Sydney Westphal, MD

,
Sunder Mudaliar, MD
Sunder Mudaliar, MD

,
Courtney Washington, BA
Courtney Washington, BA

,
Mayer B. Davidson, MD
Mayer B. Davidson, MD

Volume 11, Issue 12

Objective: To improve the treatment of risk factors of cardiovascular disease (CVD) for older patients with diabetes after a cardiac event by using a low-literacy reminder card describing these risk factors in community settings.

Study Design: A multicenter, randomized, interventional study.

Methods: Patients aged 55 years or older with diabetes hospitalized with an acute myocardial infarction, congestive heart failure exacerbation, or unstable angina were eligible to enter the study. Control and experimental patients were recruited from 4 sites and were enrolled in the study before discharge from the hospital. Experimental subjects received education and a reminder card describing risk factors of CVD. They were instructed to discuss the risk factors described on the reminder card with their primary care physician on their first appointment after discharge. Control subjects did not receive any intervention but were given consent to be able to review their charts.

P

Results: One hundred sixty patients completed the study, 82 in the control group and 78 in the experimental group. At the end of the study there was no difference in blood pressure control, lipid levels, and glycosylated hemoglobin levels between the control and experimental patients. Aspirin use and ACE inhibitor use were found to be significantly higher in the control group (= .001 and .03, respectively).

Conclusions: Reminder cards given to patients to discuss with their primary care providers in community settings did not improve process measures of CVD risk in patients with documented CVD and diabetes. Other approaches will be needed to improve the treatment of risk factors in these high risk patients.

(Am J Manag Care. 2005;11:758-764)

The leading cause of death in people with diabetes is cardiovascular disease (CVD). The 5-year mortality rate following a myocardial infarction can be 50% higher for people with diabetes as compared to people without diabetes.1 Yet recommended cardiovascular goals in patients with diabetes are often not met.2, 3 The National Cholesterol Education Program considers diabetes a CVD equivalent and not just a risk factor.4

Goals for treating cardiovascular risk factors in patients with diabetes are a blood pressure of less than 130/80 mm Hg,5 a low-density lipoprotein cholesterol (LDL-C) level of less than 100 mg/dl,6 and use of antiplatelet therapy.7 In 2 major urban medical centers, only 3.2% of patients with diabetes met goals for blood pressure, glycosylated hemoglobin (A1C), and LDL-C levels.8 In a national sample that evaluated the newer National Health and Nutrition Examination Survey (NHANES) study (1999 -2000), only 7.3% of patients with diabetes met cardiovascular treatment goals.9 Given that CVD is the leading cause of death in this country, these 2 studies demonstrate a tremendous public health need for interventions to improve the attainment of cardiovascular treatment goals in patients with diabetes.

Meeting treatment goals for at least 1 cardiovascular risk factor is good but far from perfect. Blood pressure goals alone were met in 26.7% of patients in the study involving the 2 urban centers8 and in only 12% of persons with diabetes in the NHANES III study.10 Meeting lipid treatment goals is extremely important to prevent or delay cardiovascular disease in patients with diabetes.11,12 Yet in patients with documented diabetes mellitus in the Lipid Treatment Assessment Project, only 41% met target LDL levels.13 Antiplatelet therapy has been shown to reduce coronary artery disease events in patients with diabetes but its use is also far from optimal.7 In 1 study,8 only 46% of patients with diabetes used antiplatelets.

Low-literacy patient education handouts have been used as a measure to increase compliance with preventive measures such as the pneumococcal vaccination,14 mammography screening,15 and diabetes process measures.16 The education handout specific for the pneumococcal vaccination increased rates of vaccination by 4 times compared with the control group (39% vs 11%).14 The mammography education handout was expected to increase screening with mammography by increasing awareness. However, the screening results in patients receiving the handout were no different than those who were given the usual verbal recommendation from a healthcare staff person at a doctor's office. In a study that used a diabetes reminder card to prompt providers to order necessary diabetes tests,16 patients would show the reminder card to the provider. The percentage of process measures (an evaluation of whether a test was ordered, such as A1C) carried out in those eligible to receive them did not differ from the control group. Interestingly, the control group achieved nearly 100% success for 2 goals by having a diabetes-specific progress note.

Despite these mostly negative results so far from studies using low-literacy educational handouts to improve medical care, much effort continues to be expended in creating and using them. The goal of this study was to assess whether treatment of risk factors for CVD among older patients improved as a result of using a low-literacy reminder card describing these risk factors. The reminder card was given to diabetic patients upon discharge from the hospital after a cardiac event. The patients were instructed to show the card to their primary care provider on their first follow-up visit. We wanted to determine whether the reminder card had an effect on care for high-risk patients with CVD.

METHODS

Patients

Recruitment took place at 4 sites: King/Drew Medical Center in Los Angeles, Calif (n = 53), Maricopa Medical Center in Phoenix, Ariz (n = 52), the Veterans Administration Hospital in San Diego, Calif (n = 27), and at Santa Barbara Cottage Hospital Healthcare Systems under the auspices of Sansum Diabetes Research Institute in Santa Barbara, Calif (n = 81). Enrollment period began September 2000 and ended December 2002. The Institutional Review Boards at each site approved the study. Control and experimental patients were recruited from each hospital.

Subjects were assigned to experimental or control groups in alternating sequence at the time of identification at each institution. Eligible patients were 55 years or older with diabetes of any type who were hospitalized with an acute myocardial infarction, congestive heart failure exacerbation, or unstable angina and who returned home after discharge (rather than to a nursing home or skilled nursing facility or other institution). Before discharge from the hospital, patients were asked to participate in the study. After signing the informed consent they were asked to identify the primary physician caring for their diabetes and the physician's phone number. Participants were restricted to individuals who would be followed by a primary care provider, not a cardiologist or an endocrinologist. The patient's age, sex, and hospital record number were recorded at this time.

The primary care physicians on staff at the participating hospitals received a letter advising them that the American Diabetes Association was setting up a project on diabetes education in patients previously hospitalized for a cardiac event. They were informed that 1 or more of their patients may be included in the study and that the patient's chart would be reviewed.

Intervention

The patients received a brochure explaining the relationship between heart disease and diabetes and were told about risk factors for heart disease in a brief, scripted discussion. They were also given a reminder card to take to their physician. The card described what they should ask their doctor regarding the treatment of risk factors for their heart disease, blood pressure goals, lipid levels, A1C levels, and use of aspirin (Figure). They were told that they would be contacted by telephone after discharge from the hospital to determine if they had seen their physician to discuss their care.

Within 2 weeks of discharge, another copy of the brochure and the reminder card were sent via US mail. The mailing also advised the patient to expect a phone call from a health professional to review what the patient should ask his or her physician regarding the risk factors for heart disease. Two weeks after the brochure and reminder card were mailed, the individuals were called to determine if they had received the mailing and to ask if they had seen their physician. If not, the research assistant reviewed with the patient what to ask the doctor regarding risk factors for heart disease. The patient was also encouraged to make an appointment to see his or her doctor. Every month thereafter, up to and including the sixth post-discharge month, the participants in the experimental group received follow-up phone calls and were mailed a reminder card until it was known that they had been seen by their physician.

Patients in the control group were not given the brochure or reminder card but an informed consent was given so as to review their charts 6 months later for comparison to the intervention group. No further contact was made with the patients.

Outcome Measures

The physicians' charts for the experimental and control patients were abstracted for the following CVD-related data at baseline and 6 months later: blood pressure, lipid levels, A1C level, aspirin use (or evidence of contraindications for its use or use of other antiplatelet medications). These primary outcomes were captured based on the chart recording of laboratory results or medication recording in physician notes.

Secondary outcomes were whether the patient was taking an angiotensin-converting enzyme (ACE) inhibitor (or an angiotensin receptor blocker in those with proteinuria or hypertension) and, if not, whether hyperkalemia or angioneurotic edema might have been a contraindication to their use; whether patients with triglyceride levels of =400 mg/dL or higher were taking a fibrate (as recommended by the American Diabetes Association at the initiation of the study); and whether patients with LDL-C concentrations of =100 mg/dL or higher were taking a statin.

Statistical Analysis

The sample size for this study was calculated as follows. Four quantitative end points were considered: A1C, systolic and diastolic blood pressure, LDL cholesterol concentrations, and aspirin use. Power calculation for each end point was carried out and we used the highest number of patients required for a 10-mm Hg difference in systolic blood pressure, a 5-mm Hg difference in diastolic blood pressure, or a 20-mg/dL difference in LDL concentrations between the experimental and control patients. The baseline data for the calculation were obtained from nearly 300 African American and Hispanic patients with diabetes followed in the Los Angeles County Clinics.

Calculations were performed to yield an a of less than 0.05 with a power of 80% using a 1-tailed test because our hypothesis was to determine if this reminder card improved the stated CVD risk factors in the intervention group.

The mean systolic blood pressure (±SD) was 140 ± 26 mm Hg. One hundred forty patients were necessary to show a difference of 10 mm Hg to a systolic blood pressure of 130 ± 21 mm Hg. For diastolic blood pressure, the mean ± SD was 76 ± 15 mm Hg. One hundred sixty patients were necessary to show a difference of 5 mm Hg to a diastolic blood pressure of 71 ± 10 mm Hg. For LDL cholesterol concentrations, the mean ± SD was 128 ± 44 mg/dL. A total of 106 patients would be necessary to show a difference of 20 mg/dL to a concentration of 108 ± 38 mg/dL. At first glance, 160 patients would be needed. However, this number was increased by 25% to account for the cluster effect of using multiple hospitals, the death rate, and patients moving within 6 months of discharge. Therefore, the goal was to recruit 200 patients, approximately 50 from each site.

t

Comparisons of blood pressure, lipids, and aspirin use between the groups of patients receiving and not receiving the reminder card were performed by Chi square analysis. An unpaired test was used to test for differences in A1C levels. Significance was accepted at 0.05 (1-tailed). Appropriate corrections were used because of multiple comparisons (ie, 4 end points). Data are presented as means ± SD.

RESULTS

A total of 213 patients were randomized (Table 1), 103 in the control group and 110 in the experimental group (ie, those receiving the reminder card). Ten patients had an intervention (coronary bypass graft or stent placement) after they were randomized; therefore they were not included. A total of 160 patients completed the study: 82 in the control group and 78 in the experimental group (Table 2). Of the 53 patients who did not finish the study, 14 died, 2 went to nursing homes after discharge, and 37 were lost to follow-up, either because their charts could not be located or they did not follow-up with their physician. Approximately 70 different primary care physicians were contacted. No baseline demographic differences were noted between patients who did or did not complete the study (Table 1).

No baseline differences between the control group and the experimental group were noted in individuals who completed the study although lipid and A1C levels were not measured in all patients (Table 2).

P

At the end of the study no difference was noted in end points between the control and experimental patients except for aspirin use and ACE inhibitor use, which was higher in the control group (= .001 and .03, respectively) (Table 3).

DISCUSSION

Effective therapies for reducing risk factors in persons with CVD and diabetes are extremely underutilized.2,3,8,9 In the conventional acute care model, the usual approach is to educate physicians about the results of clinical trials and to urge the rapid and broad adoption of new practices. Physicians generally are given reminders about meeting a particular disease guideline, and patients are educated only when they come to see the doctor after they are ill. Meeting evidence-based standards for important chronic public health diseases in community settings has been extremely difficult with this model, however. Despite current educational efforts, the goals for many effective therapies in different diseases are not being met. This problem is especially true for CVD in patients with diabetes.

Quality assurance efforts utilizing both process and outcome measures have been used to evaluate disease management practices for many diseases.17-19 A process measure is an evaluation of whether a test or examination was carried out. In the case of CVD and diabetes, process measures would be the measurement of A1C, fasting lipid panel, blood pressure, and the use of aspirin. The outcome measures would be the actual values of A1C, lipids, and blood pressure.

Investigators have conducted several interventional studies to try to improve CVD risk factors. In 2003, the Hastening the Effective Application of Research through Technology (HEART) trial demonstrated that implementation of a program to improve quality of CVD care was not successful.20 This study was a randomized, controlled, practice-based trial designed to evaluate an intervention to improve the quality of care provided to patients with clinically diagnosed CVD in a managed care setting. The intervention included dissemination of the American Heart Association's (AHA) recommendations for the secondary prevention of CVD, peer-comparison performance audit, and patient-specific chart reminders. The specific chart reminders were cards with the AHA's recommendations posted on the charts of patients with CVD. The study included 184 practices with a total of 605 physicians. Ninety-seven practices were randomized to intervention and 87 were randomized to the control groups. The filling of prescriptions of statins, ß-blockers, and ACE inhibitors was observed over a 3-year period. The first year was the baseline, the second year had the intervention, and the third year was the maintenance. No difference was found in any of the years in the filling of prescriptions. The lack of effectiveness in this study was attributed to the fact that the responsibility for implementation was placed on the individual physician rather than study personnel overseeing the study or intervention process.

P

In another recent study investigators reviewed ß-blocker usage in patients after acute myocardial infarction (AMI).21 The investigators examined Medicaid patients and the use of ß-blocker by their primary care providers. The intervention included mailing a letter and an informative newsletter to the physicians stating the importance of ß-blocker use in post-AMI patients. The investigators found a nonsignificant approximate 8% increase in ß-blocker prescriptions during the intervention period (<.05). The 8% increase was not sustained in the postintervention period.

These 2 studies illustrate the difficulty in improving evidence-based management of CVD and its risk factors. Other possible ways to improve outcome measures are to use case-managers and nurse-run clinics. These forms of healthcare delivery have been found to be clinically effective22,23 while being cost effective in CVD care.22 These approaches have been adapted in some large managed care settings,22 but in community settings they have met some resistance.

The present study illustrated that reminder cards given to patients to show their primary care providers in community settings were insufficient to change physician behavior and did not improve outcome measures in patients with documented CVD and diabetes. Therefore, other means of improving care have to be created to use in the current system.

Another possible mechanism to improve both process and outcome measures in CVD and diabetes care is the use of disease-specific progress notes or flow sheets, on which the provider is required to document the measurement or examination. A study that was unsuccessful in showing that reminder cards improved diabetes care process measures found that the control group that used diabetes-specific progress notes containing foot examination and A1C levels, had these 2 process measures documented almost 100% of the time.16 However, another study that evaluated several means of improving quality of diabetes care did not confirm this success.24 Twenty-nine percent of physician practices used diabetes flow sheets and 33% of the offices allocated diabetes care responsibilities to nonphysician personnel; the rest used no specific tools. Neither the flow sheets nor nonphysician personnel conferred better performance in meeting diabetes care guidelines than offices that did not use these tools.24 Therefore, it may be that flow sheets for providers may be effective in certain settings such as managed care or county facilities and not in the private sector.

The present study had several limitations. Socioeconomic status, comorbid conditions, duration of diabetes, and health insurance were all potential confounders of the intervention and were not controlled for to determine an independent association of the card. The type of insurance held by the participants ranged from preferred provider organization to none, but because this information was not documented formally, insurance assignment could not be evaluated. Baseline medications were not recorded. Had they been recorded, the results might have been more conclusive because of the knowledge that the intervention and control groups began with the same number of prescriptions.

It has been very difficult to find interventions in community settings that increase process measures, much less outcome measures of diabetes and CVD care. At a time when evidence-based medicine is transforming medical practice, there needs to be unique and possibly individualized programs not only to inform physicians, but more importantly to change physician practice behaviors.

From the Charles R. Drew University School of Medicine and Science, Clinical Center for Research Excellence, University of California, Los Angeles (DE, MD, CW, MBD); Sansum Diabetes Research Institute, Santa Barbara, Calif (LJ, SW); Department of Medicine, Maricopa Medical Center, Phoenix, Ariz (SW); Diabetes/Metabolism Section, VA San Diego Healthcare System, San Diego, Calif (SM); and the Department of Medicine, University of California San Diego (SM).

Financial support for this study was provided by the American Diabetes Association, Alexandria, Va.

Address correspondence to: Diana Echeverry, MD, Charles R. Drew University School of Medicine and Science, Clinical Center for Research Excellence, University of California, Los Angeles, 1731 E. 120th Street, Los Angeles, CA 90059. E-mail: diecheve@cdrewu.edu.