McBride P, Schrott HG, Plane MB, Underbakke G, Brown RL. Primary Care Practice Adherence to National Cholesterol Education Program Guidelines for Patients With Coronary Heart Disease. Arch Intern Med. 1998;158(11):1238-1244. doi:10.1001/archinte.158.11.1238
Copyright 1998 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1998
Clinical trials demonstrate significant benefit from cholesterol management for patients with cardiovascular disease (CVD). National guidelines recommending goals for screening and treatment were published in 1993 and widely disseminated. This study examines cholesterol screening and management by primary care physicians after the guidelines were released.
Medical records and patient surveys provided data for 603 patients with CVD, aged 27 to 70 years, from 45 practices in 4 states during 1993 to 1995. Physician surveys measured estimated performance and other variables. Physician and patient factors associated with adherence, or lack of adherence, to national guidelines were examined using univariate and multivariate analyses.
A total of 199 patients (33%) with CVD were not screened with lipid panels, 271 patients (45%) were not receiving dietary counseling, and 404 (67%) were not receiving cholesterol medication. Only 84 patients (14%) with CVD had achieved the recommended low-density lipoprotein level of less than 2.58 mmol/L (100 mg/dL) and 302 (50%) had triglyceride levels lower than 2.26 mmol/L (200 mg/dL). Patients with a revascularization history and higher low-density lipoprotein and/or triglyceride levels were more likely to receive treatment, but other patient factors, including CVD risk factors, did not predict treatment. Physician specialty was not associated with differences in treatment, but physicians in practice for fewer years ordered more lipid panels.
Most patients with CVD in primary care were not receiving cholesterol screening and management as recommended by the National Cholesterol Education Program guidelines in the 2 years after their release. Increasing cholesterol screening and treatment should be a priority for practice quality improvement and could result in significant reductions in CVD events for high-risk patients.
MANY OF the recurrent cardiovascular disease (CVD) events that lead to high short-term morbidity and mortality for patients with CVD can be prevented or delayed with management of cholesterol levels.1- 5 The Adult Treatment Panel (ATP-II) of the National Cholesterol Education Program published guidelines in 1993 that placed a strong emphasis on secondary prevention with a recommendation to manage cholesterol in those with known CVD.6 This recommendation is reinforced by clinical trials that demonstrate a reduction in the morbidity and mortality rates of 30% to 50% with management of cholesterol levels for patients with CVD.1,2,5,7 Cholesterol management also results in favorable cost-benefit outcomes with the resulting reduction in procedures and hospitalizations related to CVD.8
Based on this evidence, it is imperative to evaluate and treat abnormal cholesterol levels in persons with CVD. The ATP-II guidelines recommend that patients with any manifestation of CVD, including coronary heart disease, cerebrovascular disease, or peripheral vascular disease, should have their low-density lipoprotein (LDL) level reduced to less than 2.58 mmol/L (100 mg/dL) and triglyceride (TG) level lowered to less than 2.26 mmol/L (200 mg/dL).6 Because LDL was the primary treatment target in the trials, it was emphasized in the ATP-II recommendations. A majority of patients with CVD will require medication in addition to lifestyle changes to reach this primary goal of an LDL level lower than 2.58 mmol/L (100 mg/dL).5
Recent studies suggest that physician management of high blood cholesterol levels is suboptimal for both primary care physicians9- 11 and cardiologists.12 This evidence shows that less than 20% to 30% of patients with CVD in the United States are receiving cholesterol-lowering therapy, but these few studies were conducted in single academic practices, looked only at single visits, or examined practice data prior to the ATP-II guideline release.9- 12
The Health Education and Research Trial (HEART) examined heart disease prevention services in primary care practices for a 5-year period. Baseline practice data from HEART provide an opportunity to (1) evaluate primary care physician management of cholesterol for their patients with CVD; (2) measure how many primary care patients with CVD are meeting the ATP-II treatment goals; and (3) examine variables that predict cholesterol level management in primary care practices. We report herein on dietary and pharmacological therapy provided for patients with CVD in this diverse group of primary care practices, their level of success in meeting ATP-II treatment goals prior to HEART clinical trial interventions, and physician and patient factors that influence cholesterol screening and management.
Primary care practices were recruited from within a 160-km radius of each of 4 midwestern regional centers: Madison, Wis; Minneapolis, Minn; Iowa City, Iowa; and Eau Claire, Wis. We have previously described HEART recruitment and methods in detail.13,14 Eligible practices had a majority of primary care physicians (defined as internists, family physicians, or general practitioners); could not be a residency or university practice; and had no prior participation in preventive service trials. Practices could be affiliated with larger health care organizations, including hospitals and health maintenance organizations. Eligible physicians had been in primary care practice at that location for at least half-time for 1 year or more and planned to stay. Of the 86 eligible practices, 52 (60%) consented to participate and 45 practices (52%) were enrolled, from which 159 physicians participated.
The University of Wisconsin Committee for the Protection of Human Subjects, Madison, and the institutional review boards of the collaborating centers reviewed and approved the study methods. Patients signed informed consent forms prior to participation to allow an audit of their medical records. Eligible patients included those aged 21 to 70 years without a diagnosis of cancer, terminal illness, or recent major surgery, and with at least 2 practice visits in the prior 2 years. There were 5333 patient participants in HEART at baseline for an average of 34 patients per physician. Data collection from medical records and patient and physician questionnaires occurred from August 1993 (Madison) to February 1995 (Eau Claire). The selection of the 603 patients for this study was based on a medical record diagnosis of CVD, including prior myocardial infarction, coronary artery bypass grafting or percutaneous transluminal coronary angioplasty, peripheral vascular disease, or a diagnosis of angina pectoris.
Patient questionnaires were mailed to all eligible patients using a previously described protocol.14 The patient questionnaire included demographic data, family and personal history of CVD, and personal coronary risk factors, including hypertension, smoking, diabetes mellitus, and exercise level. Patients reported whether their cholesterol levels had been measured and they had been told their results by clinic staff or physicians, and whether they were advised to change their diet and/or were prescribed medication for management of cholesterol levels.
Physician questionnaires asked providers for their attitudes, beliefs, and estimates of practice behavior regarding preventive services. All 159 participating physicians completed questionnaires reporting their beliefs about cholesterol-related coronary risk factors, patients' desire to know their cholesterol levels, and patient receptiveness to dietary or exercise advice. They answered detailed questions about their practice routines regarding cholesterol screening and management of elevated cholesterol levels for primary and secondary prevention of CVD.
The medical record review included the following information: status of patients with CVD, the presence of family history of CVD, diagnosis and treatment of hypertension, smoking status and its treatment, diagnosis of diabetes, body mass index (measured as the weight in kilograms divided by the square of the height in meters), and lipid levels recorded within the last 5 years, including total cholesterol, high-density lipoprotein cholesterol (HDL), and TG. We also noted recorded cholesterol diet management and exercise advice, cholesterol medication, information on weight and height, and general diet and exercise information. Lipid measurements were those values found in the medical record.
We computed LDL levels using the Friedewald equation for total cholesterol, HDL, and TG levels with the same date for 54% of records. When patients did not have total cholesterol, TG, and HDL levels measured on the same date, we computed an estimated LDL level using the patients' most recent HDL level (an additional 13% of records), assuming physicians used this information for clinical decisions and that the HDL level is relatively stable over time. We thus have an estimate of LDL levels for 67% of the patients with CVD in the database.
The data collection strategy provided multiple information sources for a comprehensive evaluation of physicians' screening and management of cholesterol levels. We have demonstrated that the medical record is accurate for medical diagnoses, laboratory test results, and prescriptions, while information on counseling activities is most complete when collected from patient surveys.15 The data presented herein on counseling were collected from the patient questionnaires and the laboratory values and medication prescription information from the medical record audit.
Computer software (SPSS Inc, Chicago, Ill) was used to generate general descriptive statistics and to test univariate associations between selected variables with χ2 statistics, t test, and 1-way analysis of variance where appropriate. Logistic regression analysis and the hierarchical logistic regression analyses were performed with SAS version 6.07 (SAS Institute Inc, Cary, NC). The BMDP program16 was used to generate the ordered polytomous logistic regressions.17
The physicians, mainly family physicians, had been in practice an average of 13 years and were predominantly male (Table 1). More than 85% reported that they believe that LDL and HDL cholesterol levels have a moderate to large effect on CVD. All physicians reported that they believe their patients want to know their cholesterol levels, 80% believe that diet can be effective in controlling cholesterol levels, and 89% stated that their patients are receptive to a physician's dietary advice. They reported that they initiate diet therapy at a mean LDL level of 3.31 mmol/L (128 mg/dL) and medication at an average LDL level of 3.46 mmol/L (134 mg/dL) for their patients with CVD. When looking at physician responses by specific LDL level, only 20% recommend dietary therapy for patients with CVD at an LDL level of 2.58 mmol/L (100 mg/dL), while 70% reported they initiate diet therapy when a patient's LDL level has reached 3.36 mmol/L (130 mg/dL), and 9% wait until the LDL level is 4.13 mmol/L (160 mg/dL) or higher. Sixteen percent stated that they will prescribe medication after a trial of dietary therapy if a patient's LDL level is higher than 2.58 mmol/L (100 mg/dL), an additional 54% will prescribe medication if the LDL level is 3.36 mmol/L (130 mg/dL) or higher, and 18% wait until the LDL level is higher than 4.13 mmol/L (160 mg/dL).
Physicians estimated that they obtain a lipid panel for 86% of their patients with CVD, a diet history for 63%, provide dietary advice to 81%, and prescribe medication to 42% of these patients. They reported scheduling follow-up visits more than 80% of the time and referring about 25% of their patients to practice staff members or outside cholesterol management programs. Eighty-six percent were satisfied with their clinic's procedures for cholesterol screening and 71% with patient cholesterol follow-up services, but only 49% were satisfied with their practice's diet counseling services for patients with high cholesterol levels.
Comparing survey data across primary care specialties did not demonstrate significant differences in attitudes or estimates of preventive services. There were no significant differences between internists or family practice physicians for lipid panel screening, dietary counseling, medications prescribed, the percentage of patients who reached a goal of an LDL level lower than 2.58 mmol/L (100 mg/dL), or patients reaching a goal of a TG level lower than 2.26 mmol/L (200 mg/dL).
Of the 5333 patients in the HEART baseline sample, 603 (11%) had medical record documentation of CVD. Table 2 lists the characteristics of these patients by sex. Although most patients with CVD were older than 50 years, their ages ranged from 27 to 70 years. Most patients were white, reflecting the demographics of the region. A total of 241 patients (40%) had some college education, and more men were married. A history of smoking was reported by 410 (68%) of these patients, while few reported smoking in the past 2 years or were currently smoking. They were more likely to have a diagnosis of high blood pressure and diabetes than the general primary care patient population. Current regular exercise 2 or more times per week was reported by 332 (55%) of the patients with CVD, while 157 patients (26%) reported no regular physical activity. Body mass index was 27 to 30 kg/m2 for 28% of patients and greater than 30 kg/m2 for 45%.
The medical record review found total cholesterol levels available for 579 patients (96%); however, 75 (13%) of these at-risk patients reported that they were unaware their total cholesterol values, and 38 members of this group had a cholesterol level higher than 6.20 mmol/L (240 mg/dL) in their medical record. Lipid panels (with an HDL and TG level to determine LDL values) were not available in the records of 204 (33%) of the patients with CVD (Table 3). A total of 543 patients (90%) had a TG level recorded in the past 5 years, and only 452 (75%) had any HDL measurement in the chart within the last 5 years. No sex differences among patients were noted in screening for total cholesterol or TG levels; however, the HDL levels necessary for a lipid panel were significantly more likely to be drawn in male than female patients with CVD (77% vs 67%; P<.01). Of those patients for whom we were able to compute an LDL level, 215 (54%) had LDL levels higher than 2.58 mmol/L (130 mg/dL); 128 (32%) had LDL levels of 2.58 to 3.34 mmol/L (100-129 mg/dL); 124 (31%) had HDL levels lower than 0.90 mmol/L (35 mg/dL); and 239 (60%) had TG levels higher than 2.26 mmol/L (200 mg/dL). Because some lipid level abnormalities occur in combination, we found that 73% of these patients with CVD with a lipid panel had at least 1 abnormal cholesterol reading.
Only 56 patients (14%) with a computed lipid panel available had an LDL level lower than 2.58 mmol/L (100 mg/dL) and 200 (50%) were at a goal TG level lower than 2.26 mmol/L (200 mg/dL) (Table 1). Male patients were no more likely to have achieved an LDL level lower than 2.58 mmol/L (100 mg/dL) than female patients, and physicians of both specialty groups had the same percentage of patients reaching their goal LDL levels. Diabetes or current smoking did not increase the likelihood that a patient received cholesterol-lowering medication, while the presence of hypertension was associated with increased prescription of cholesterol-lowering medication (36% vs 26%; P<.05).
The treatment patients received as it related to their LDL and TG levels is listed in Table 3. Patients who had no LDL level recorded were less likely to report cholesterol diet advice and less likely to have a cholesterol-lowering medication prescription found in their medical record. Dietary counseling and prescriptions of cholesterol-lowering medication increase as LDL values increase. Cholesterol-lowering medication was prescribed for one third of patients (Table 3), but more often if the LDL level was higher than 3.36 mmol/L (130 mg/dL) and the TG level was higher than 2.26 mmol/L (200 mg/dL). Patients with HDL levels lower than 0.90 mmol/L (35 mg/dL) were more likely to report diet advice and significantly more likely to have medication prescribed than those with a higher HDL level. When examining patients who had been diagnosed as having CVD, those who had undergone a revascularization procedure (coronary artery bypass grafting or percutaneous transluminal coronary angioplasty) were the most likely to receive dietary counseling or cholesterol-lowering medication, while those who were diagnosed as having angina pectoris as the only CVD were the least likely to receive therapy.
The types of cholesterol-lowering medications prescribed for these patients are shown in Table 4. The most frequently used medications were 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors, which were prescribed nearly twice as often as fibrates or niacin whether or not the TG level was elevated. Bile acid sequestrants and psyllium husk were prescribed to fewer patients overall than the other medications. If TG levels were higher than 2.26 mmol/L (200 mg/dL) or HDL levels were lower than 0.90 mmol/L (35 mg/dL), the physicians were more likely to prescribe fibrates or niacin, but the 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors were still the most prescribed medications for all subsets of patients.
We examined the data for geographic differentials and differences that might show that dissemination of national guidelines had an impact on cholesterol screening or management over time. There were no significant differences between geographic regions in total cholesterol or TG level screening, diet advice, or the prescription of cholesterol-lowering medications. Minnesota practices were significantly more likely to have an HDL level recorded in the patients' medical record and to recommend exercise for these patients with CVD (P<.001). There were no significant differences between the first region measured and the last region, suggesting that timing related to guidelines was not a factor.
Logistic regression analysis was performed to provide a test of the relationship of physician and patient factors that may have contributed to the provision of dietary counseling and medication therapy. This analysis supported the univariate associations previously discussed. Physicians' sex and specialty had no significant relationship to lipid panel screening, reported diet advice, or cholesterol-lowering medication prescriptions. Physicians in practice longer were significantly less likely to order a lipid panel than those who entered practice more recently; however, once they ordered a lipid panel, their diet advice and cholesterol prescriptions were similar.
Patient factors that predicted screening included a higher level of education, coronary artery bypass grafting, and a higher total cholesterol level. The logistic regression analysis also showed that an elevated total cholesterol level and the presence of a lipid panel in the medical record were significantly associated with lipid dietary and pharmaceutical management. Similar associations were noted if LDL or HDL values replaced total cholesterol in the equation. Patients having had coronary artery bypass grafting procedures or percutaneous transluminal coronary angioplasty were almost twice as likely to report diet advice and more than twice as likely to have medication prescribed. Patients diagnosed as having congestive heart failure were only half as likely to report treatment. Patients with Medicare or Medicaid coverage compared with those who had private insurance were more likely to have a lipid panel ordered and reported diet advice, but no difference was noted for medication prescription. The logistic regression analyses showed no relationship between patient age and sex, number of visits, or even the presence of an additional CVD risk factor and provision of cholesterol management services. Adding a patient/sex interaction term did not reveal any additional association.
This study of a wide range of community-based primary care practices indicates that primary care physicians did not consistently follow the recommendations of the second National Cholesterol Education Program ATP-II6 in the 2 years after the release of the guidelines. Many high-risk patients with CVD did not receive lipid panels, were not treated, and even fewer reached the recommended LDL goal level of less than 2.58 mmol/L (100 mg/dL) and TG goal level of less than 2.26 mmol/L (200 mg/dL). Patients in these practices clearly need more complete screening, treatment, and follow-up to achieve these important goals. Physician survey responses in this study indicate that they may not have been aware of or agree with the specific LDL and TG goal levels for patients with CVD recommended in the ATP-II, although they report being in favor of treating cholesterol in these patients. Physicians overestimate the treatment provided to patients with CVD, indicating that practice data feedback and quality improvement efforts may be helpful to improve the cholesterol screening and treatment of these patients.
Our study is consistent with a recently published evaluation of the National Ambulatory Medical Care Surveys, a national database that reported data from 56215 visits to 2332 physicians in 1991 and 1992.10 The results indicated that while counseling was more likely for patients with CVD, cholesterol testing and therapy did not occur often enough in patients with CVD. Unlike the National Ambulatory Medical Care Surveys study, we did not find significant differences between family physicians and internists. The data from that study are limited by (1) the use of single-visit data that are either the first physician visit or a general medical examination rather than all physician visits collected over a period; and (2) the collection of the data before the release of the 1993 guidelines and the publication of many clinical trials supporting aggressive cholesterol therapy for patients with CVD.10
Our findings extend those of the National Ambulatory Medical Care Surveys study10 by including data obtained after the release of the ATP-II guidelines, providing data recorded in the medical record for the previous 5 years, and including more detail on the physicians and patients. With more comprehensive data to evaluate we noted more screening and treatment, including lipid panel measurement and dietary counseling, than other published reports that evaluated fewer visits or only visits before the 1993 guidelines.9- 11 We have described barriers to cholesterol services in practice,18,19 but with the numerous clinical opportunities available for patients in this study (average visits, >7 per patient) and the importance of cholesterol management to reduce CVD risk, these barriers must be overcome. It is clear that in this study and nationwide there is a significant gap in the primary care screening and treatment of cholesterol disorders in patients with CVD.10,11
Our study data suggest that a number of factors may be related to the lack of adherence to national guidelines and can be the focus of future education and research efforts: (1) many physicians were either unaware of or did not agree with the ATP-II guidelines that were released prior to the study that recommended lipid panels in patients with CVD and a goal LDL level lower than 2.58 mmol/L (100 mg/dL); and (2) patients with a past revascularization and very high total cholesterol, LDL, and TG levels receive more counseling and therapy. These findings suggest that physicians are more aggressive when a patient is thought to be at higher risk and that physicians may not perceive a benefit to assist all patients with CVD to reach a goal LDL level lower than 2.58 mmol/L (100 mg/dL).
Limitations to the generalizability of our findings include a self-selected population of physicians and patients. Because this was a clinical trial, consent was necessary for practice, physician, and patient participation. However, these practices included a wide range of settings and the findings are consistent with reports in other settings.9- 11 If selection bias were an issue, it is likely that screening and treatment would have been even lower in other nonparticipating practices because participation should have predicted a greater interest in prevention of heart disease. Timing related to the guidelines' release or new research publications did not appear to affect these results because we saw no significant difference from practices studied from 1993 to 1995. The differences in screening and treatment found in Minnesota practices, most of which were affiliated with managed care organizations, suggests that managed care affiliation may influence care, but this finding would require more research. Because the dissemination of guidelines and trials can be a delayed process, timing of the study could be important. However, our analysis showed no temporal effect over 2 years since practices evaluated nearly 2 years apart showed no significant differences in screening and treatment. Our methods of measurement could only judge screening and counseling provided at the primary care practice and not at other locations, although recorded treatment with medications is likely to be accurate because this treatment would have been recorded in the primary care practice.
National and local strategies are necessary to improve the low rates of adherence to national prevention guidelines by primary care physicians and to improve the quality of care provided to patients with CVD. Treatment of cholesterol is an effective method for reducing CVD events, and the benefits may be even greater for women, older patients, and those with other risk factors including smoking.1,2,5 With consistent findings that lowering LDL levels to below 2.58 mmol/L (100 mg/dL) reduces the numbers of recurrent CVD events and symptoms and that the benefits of treatment accrue soon after initiating treatment, most patients should be screened and considered for aggressive lifestyle changes and medical therapy.1- 6 It is essential that primary care physicians develop strategies to test and treat all eligible patients with CVD or to document in which cases exceptions should be made.
Our results suggest that a number of strategies are needed to improve the quality of cholesterol management in patients with CVD. Educational programs for physicians must emphasize the importance of cholesterol management (and other secondary prevention treatments) for patients with CVD and point out the differences in event reduction for higher risk patients. The consistent findings between practices, the failure of physicians to reach their own standards, and prior research on physician performance suggest that practice system changes and quality improvement efforts will be needed to improve adherence to national guidelines.20 Because many patients are not tested, followed up consistently, or treated when it is appropriate, the institution of practice systems, such as reminders and prompts, practice staff support, and case management strategies, needs to be considered. Quality assurance and quality improvement efforts should emphasize cholesterol screening and treatment, especially for patients with CVD, because of the inappropriately low rates of screening and treatment noted in primary care studies and the large potential benefits.1,2,5,7,8
Accepted for publication November 4, 1997.
This research was supported by Public Health Services grant RO1 HL-47554 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md.
Presented in part at the Fourth International Symposium on Preventive Cardiology, Montreal, Quebec, July 1, 1997.
We thank the HEART practice physicians, staff, and patients for their willingness to participate and the Wisconsin Research Network, Madison, and Health Partners, Minneapolis, for their assistance.
Reprints: Patrick E. McBride, MD, MPH, HEART, 777 S Mills St, Madison, WI 53715-1849 (e-mail: firstname.lastname@example.org).