- Email: [email protected]
Effects of Practice Setting on Quality of Lipid-Lowering Management in Patients With Coronary Artery Disease
Effects of Practice Setting on Quality of Lipid-Lowering Management in Patients With Coronary Artery Disease David J. Harnick, MD, Joel L. Cohen, MD, Clyde B. Schechter, Valentin Fuster, MD, PhD, and Donald A. Smith, MD, MPH
MD,
We undertook a study to determine whether there were differences in the quality of lipid management in patients with coronary artery disease (CAD) in 2 different practice settings (which represent different socioeconomic classes), and to determine the level of compliance with the National Cholesterol Education Program guidelines by academic physicians in managing patients with CAD. A retrospective cross-sectional study was performed using a systematic chart review of 270 medical records (131 from the cardiology clinic, 139 from the cardiology private practice) of patients with known CAD at an academic tertiary care center in New York City. The total proportion of patients with CAD having a lipid profile ordered in the clinic and private suite was 43%. Of these people, 22% had a low-density lipoprotein cholesterol (LDL) <100 mg/dl and 54% had an LDL <130 mg/dl (10% and 23% of the total population, respectively). The total proportion of patients taking lipid-lowering medications was 29%. When comparing the quality of treatment between the 2 settings, there were
no statistically significant differences in the percentages of patients who had lipid profiles measured (40% clinic vs 47% private suite, p >0.10), in the percentage of patients with LDL <130 mg/dl (50% clinic vs 57% private suite, p >0.10) or in the weighted percentage of patients taking lipid-lowering medications (29% clinic vs 48% private suite, p 5 0.099). The performances of individual physicians, however, varied widely. The percentages of patients with lipid profiles measured by individual physicians ranged from 0% to 83%, while the percentages of patients on drug treatment by a physician ranged between 10% and 88%. These findings indicate that socioeconomic differences, represented by different practice settings, do not account for differences in the screening for, control of, or use of medications in managing hyperlipidemia. Rather, individual physicians are accountable for differences in lipid management. Q1998 by Excerpta Medica, Inc. (Am J Cardiol 1998;81:1416 –1420)
everal previous studies have documented less than optimal physician compliance with regard to folS lowing the National Cholesterol Education Program
lowering treatment according to the national recommendations and if these patients were receiving a similar quality of care depending on the type of health care setting they visited.
(NCEP) guidelines for lipid management.1–3 These studies, which analyzed the issue of physicians’ attitudes and practices toward secondary prevention of CAD, produced similar results and concluded that no group of physicians reflected an aggressive cholesterol reduction attitude. One investigator found that physicians ordered lipid profiles in only 18% of their patients with CAD,1 whereas another study found that physicians were likely to begin drug treatment for hypercholesterolemia only at total cholesterol levels ranging from 296 to 319 mg/dl.2 Furthermore, they determined that physician practices vary among one another, especially between different specialties and that the management of patients with existing CAD by all specialties is not in accordance with the NCEP guidelines.2 Another trial of patients on monotherapy with HMG coenzyme A reductase inhibitors to reduce cholesterol demonstrated only 33% of patients achieving NCEP guidelines.3 In this study we examined whether this patient population was receiving lipidFrom the Cardiovascular Institute, and the Department of Community Medicine, Mount Sinai School of Medicine, The Mount Sinai Medical Center, New York, New York. Manuscript received September 22, 1997; revised manuscript received and accepted February 2, 1998. Address for reprints: Donald A. Smith, MD, MPH, Mount Sinai School of Medicine, Cardiovascular Institute, PO Box 1014, One Gustave L. Levy Place, New York, New York, 10029.
1416
©1998 by Excerpta Medica, Inc. All rights reserved.
METHODS
Setting: Our study was conducted at both the cardiology clinic and private faculty cardiology suite at the Mount Sinai Medical Center in New York City. This major urban teaching hospital is located on the upper east side of Manhattan and lies adjacent to Spanish Harlem in New York City. The 2 different practice settings that we examined differ mainly in the financial reimbursements that each receives. The private suite receives 40% of its reimbursements from private sources, whereas the clinic receives only 14% of its revenues from private sources (Table I). Similarly, Medicare reimbursements are greater in the private suite (45%) than in the clinic (31%). On the other hand, .50% of the reimbursements that the clinic receives are either from Medicaid or are nonreimbursable visits, whereas only 15% of the private suites’ income is from these sources. Thus, the reimbursement characteristics of each setting depict the different socioeconomic characteristics of the patients who attend them, with a larger proportion of Medicaid patients attending the clinic and more affluent classes attending the private faculty practice. Patients: A cross-sectional study of patients with existing CAD was performed using a chart review 0002-9149/98/$19.00 PII S0002-9149(98)00209-4
TABLE I Source and Percentage of Reimbursement by Setting Office Site (% of funding)
Reimbursement Source
Private Clinic Suite (n 5 131) (n 5 139)
Private funding Private insurance Health Maintenance Organization Self-pay Subtotal, private source Medicare Subtotal, private source 1 Medicare Medicaid Nonreimbursable visits
16 1 1 18 41 59 39 33
(12%) 38 (27%) (1%) 14 (10%) (1%) 4 (3%) (14%) 56 (40%) (31%) 62 (45%) (45%) 118 (85%) (30%) 21 (15%) (25%) 0 (0%)
method. CAD was defined by a history of 1 of 4 conditions: (1) a myocardial infarction, as documented by a physician note in the chart; (2) a coronary artery bypass graft; (3) a percutaneous transluminal coronary angioplasty; or (4) a moderately or severely positive thallium stress test, as documented in the chart by a nuclear cardiologist. In the faculty practice, subjects were selected by reviewing each physician’s charts in an alphabetical manner, looking for the first 15 patients with known CAD. In the clinic, subjects were randomly found by searching through numbered patients charts in ascending order. Patients in both settings were included in the study if they fulfilled the following criteria: (1) had at least 3 outpatient visits to the same physician at the Mount Sinai Hospital between December 1993 and December 1994; and (2) met 1 of the criteria for CAD as mentioned above. In some cases, 15 subjects could not be found for each physician in each setting, and thus the numbers of patients varied by physician and setting. The hospital patient number, gender, and present age of each of the patients meeting these criteria were recorded. The charts of these patients were then reviewed, focusing particularly on 2 factors: (1) if a lipid profile had been ordered within the last 12 months, and (2) if the patient had currently been prescribed a lipid-lowering pharmacologic agent. To ascertain the presence and values of a lipid profile ordered through the hospital laboratory, the hospital patient number was used to reference the hospital laboratory test computer. This was a necessary step as physicians often failed to record the lipid profile order within the charts. Also, many of the charts did not include the most recent lab profiles, so the current lipid values for patients needed to be obtained by the computer. Physicians: Physicians who participated in this study are salaried faculty at the Mount Sinai Hospital and have patients in both the cardiology clinic and the private faculty cardiology suite. In addition to their base salary, the physicians earn a small percentage of their total income generated in the faculty practice. Lipid profiles: The lipid profiles obtained in Mount Sinai Hospital were performed in the following man-
ner: (1) Total cholesterol, high-density lipoprotein (HDL) cholesterol and triglyceride measurements were obtained on the hospital’s Technicon Chem 1 system (Bayer Diagnostics, Tarrytown, New York) according to the manufacturer’s recommendation. (2) Low-density lipoprotein (LDL) cholesterol was calculated using the Friedwald formula.5 Quality control is monitored through the use of bilevel controls run 3 times daily plus periodic surveys conducted by the American College of Pathologists and the New York State Department of Health. Lipid profiles performed within the hospital accounted for most of the tests in both settings. Some lipid profiles of the patients attending the private cardiology suite, however, were performed by outside clinical laboratories. Data analysis: Data were entered into a Lotus 1-2-3 spread sheet, from which simple descriptive statistics were generated. Hypothesis tests were calculated using SYSTAT6 version 5.0 for DOS. Crude comparisons between groups on continuous variables were made using the Student t test, and for categorical variables using the Pearson chi-square and MantelHaenszel chi-square statistics. Linear regression was used for adjusted comparisons of continuous variables across groups, and logistic regression was used for adjusted comparison of categorical variables across groups.
RESULTS
Demographic information: PATIENT PROFILE: We reviewed 270 patient charts from the Mount Sinai Hospital. One hundred thirty-one were obtained from the cardiology clinic and 139 from the private practice suite. All patients were previously diagnosed with CAD as stated in the Methods section. The age (mean 6 SD) of clinic and of the private suite patients was 61 6 10 and 65 6 11 years, respectively (p ,0.003). In the clinic, nearly equal numbers of charts of men and women were reviewed (49%: 51%), whereas more charts of men were reviewed in the private suite (66%:34%). Of the patients who had an LDL measured (n 5 117), clinic patients had a mean LDL of 133 mg/dl and private patients had a mean LDL of 132 mg/dl (p .0.10). Because LDL values were positively skewed, a linear regression analysis using log LDL values revealed the geometric mean LDL cholesterol level to be 1.004 (95% confidence interval 0.88 to 1.14) times as high in the clinic versus the private practice setting, adjusting for age, gender, and individual physician. The power of the model to detect a 5% difference in the LDL cholesterol level by site was 80%. In contrast, women (n 5 55) had a higher mean LDL (145 6 48 mg/dl) than men (n 5 66, 122 6 38 mg/dl), (p ,0.02), a significant difference that persisted when adjusted for differences in age, clinic versus private setting, and the treating physician. A further analysis that showed a higher proportion of women who had an LDL .100 mg/dl (p ,0.05) confirmed this trend. However, the percentage of women with an LDL .100 was no longer statistically CORONARY ARTERY DISEASE/LIPID MANAGEMENT
1417
TABLE II Lipid Management Results by Setting
Lipid profiles in last 12 months LDL cholesterol #130 mg/dl LDL cholesterol #100 mg/dl On lipid-lowering medications Mantel Haenszel chi-square (weighted statistic) LDL cholesterol .130 mg/dl on lipid-lowering medications
Overall (n 5 270)
Clinic (n 5 131)
117 63 26 77
52 26 11 29
(43%) (54%) (22%) (29%)
26 (50%)
significant after applying the Mantel-Haenszel stratification analysis to control for the physician (p 5 0.09). Thus, although overall there are more women than men likely to have an LDL .100, this apparently happens because physicians who are less likely to achieve an LDL #100 mg/dl are treating more of the women. PHYSICIAN PROFILE: The physicians participating in our study are attending physicians at the Mount Sinai Medical Center who see patients in both the clinic and the private practice suite. A total of 11 physicians participated in our study. Ten physicians are cardiologists and 1 is an endocrinologist/lipids specialist with additional expertise in public health. Their years in practice range from 1 to 15 (average 6.3). PHYSICIAN PRACTICE PATTERNS: The overall physician performance was compared with the recommended 1993 NCEP guidelines for managing hypercholesterolemia in patients with CAD. The following parameters were examined: (1) presence of a lipid profile within the past 12 months; (2) percentage of patients with LDL ,100 mg/dl; (3) percentage of patients with LDL ,130 mg/dl; (4) percentage of patients taking a lipid-lowering medication; and (5) percentage of patients with LDL .130 mg/dl on lipidlowering medication. Of all 270 patients, 117 (43%) had a lipid profile in the preceding 12 months. Of these patients, 26 (22%) had an LDL ,100 mg/dl and 63 (54%) had an LDL ,130 mg/dl (10% and 23% of the total population, respectively). The proportion of all patients on a lipidlowering medication was 29% (n 5 77). Among the 153 patients without a lipid profile, 26 were nevertheless receiving lipid-lowering medication. Of patients with an LDL .130 mg/dl, exactly 50% received drug treatment (Table II). When examining the frequency of ordering lipid profiles by individual physicians, considerable variability is evident between physicians, but apparently less variation within each physician’s own practice patterns (Figure 1). For example, physician A ordered lipid profiles in approximately 70% of his patients (of all patients seen), whereas physician J ordered lipid profiles only in about 10% of his (data not shown). Apparently, some physicians ordered more lipid profiles in the clinic, as seen in physicians A, D, G, and K, whereas some ordered more in the private suite (Figure 1). The trend within each physician’s ordering 1418 THE AMERICAN JOURNAL OF CARDIOLOGYT
VOL. 81
(40%) (50%) (21%) (22%)
10 (40%)
Private Suite (n 5 139) 65 37 15 48
(47%) (57%) (23%) (35%)
16 (59%)
p Value .0.10 .0.10 .0.10 ,0.05 0.099 .0.10
of lipid profiles was relatively constant between the 2 settings for each physician. Note that the lipids specialist, physician K, ordered lipid profiles more frequently than any of the cardiologists (83% of his total patients had a lipid profile, data not shown). When analyzing the rates of lipid-lowering medication prescriptions by physicians, a similar pattern was noted (Figure 2). Once again, the physicians differed widely between one another, prescribing medications to between 10% and 50% of their patients, and in general the cardiologists prescribed less medications than the lipids specialist (prescribed to 88% of patients, data not shown). Note that some physicians prescribe to more patients in the clinic and others prescribe to more in the private suite, as was the pattern with the ordering of lipid profiles. Although, there are occasional physicians for whom the difference between settings are great (physicians B, D, and H), in general they are consistent between the 2 settings (Figure 2). It is also interesting to note that physicians A, B, and C, who ordered more lipid profiles, tended to prescribe more lipid-lowering medications. COMPARISON OF CARE PROVIDED TO PATIENTS IN CLINIC AND IN THE PRIVATE SUITE: There was no statis-
tically significant difference in the proportions of patients who had lipid profiles performed (40% clinic vs 47% private suite, p .0.10), in the percentage of patients with lipid profiles who had an LDL #130 mg/dl (50% clinic vs 57% private suite, p .0.10), and in the percentage with an LDL #100 mg/dl (21% clinic vs 23% private suite, p .0.10) (Table II). The proportion of patients taking lipid-lowering medications did differ substantially in the 2 settings, however, with 22% of clinic patients taking medications compared with 35% of private practice patients taking medications (p ,0.05) (Table II). To see if this was caused by variability in numbers of patients seen by specific physicians in the 2 settings, the original 2 3 2 analysis was subdivided into 11 smaller 2 3 2 analyses (1 for each physician). The association between drug presence or absence in the clinic versus the private suite was tested by individual physician and then combined using the Mantel-Haenszel chi-square statistic. This summary statistic, which weighs each physician equally, correcting for any differences in the numbers of patients seen by each physician in each setting was not significant (chi-square 5 2.723, p 5 0.099), demonstrating no difference in the use of JUNE 15, 1998
senting the patients’ socioeconomic status further influenced the quality of care. Our results demonstrated 3 important findings. First, there were no significant differences in the management of hyperlipidemia in patients with CAD between a hospital clinic and private practice setting. Second, physicians who participated in this study did not adequately comply with the 1993 NCEP guidelines. In fact, ,50% of the study patients had a lipid profile performed within 12 months of treatment, of which 78% had an LDL cholesterol level FIGURE 1. Percentage of patients with lipid profiles by individual physician. Physicians A above the recommended level of through J: cardiologists. Physician K: endocrinologist. The numbers below the horizontal 100 mg/dl for secondary prevenaxis are the numbers of patient charts reviewed in each setting. Filled bars above each tion.7–9 Although at the time of letter represent the percentages of clinic patients by physician; open bars represent the this study a target LDL level of percentages of private patients by physician. The last column on the right represents the percentage of patients in the clinic (filled) and in the private suite (open). comb. 5 com,130 mg/dl was recommended bined. for secondary prevention, we found 46% of the study patients to have values above this goal. Last, our study revealed that lipid management varies more widely among individual physicians than between different practice settings. One possible reason for only partial compliance of specialists with the NCEP is that lipid management is left to the prerogative of the primary care physician. This is evident in our study because 17% of patients without a documented lipid profile by their cardiologist were receiving lipidlowering therapy. Previous studies have shown FIGURE 2. Percentage of patients on lipid-lowering therapy by physician. Physicians A differences in quality of care bethrough J: cardiologists. Physician K: endocrinologist. The numbers below the horizontal tween different care areas, namely axis are the numbers of patient charts reviewed in each setting. Filled bars above each letter represent the percentages of clinic patients by physician; open bars represent the Medicaid/Medicare clinics and percentages of private patients by physician. The last column on the right represents the fee-for-service private pracpercentage of patients in the clinic (filled) and in the private suite (open). comb. 5 com10 –13 The flaws in these studtices. bined. ies, which were circumvented in this study, included the use of subjective measurements,11 indrugs in the clinic or private suite. In effect, we found equality of services offered to different patients,12 and that physicians who typically prescribe lipid-lowering the use of different physicians in different settings medication less frequently (in either setting) ac- which confounded a simple setting comparison.13 Similarities in the management of lipids in 2 difcounted for more of the patients seen in the clinic setting, resulting in an apparent, but deceptive, overall ferent care areas with 2 different patient populations in finding of less medications prescribed in the clinic this study were surprising. We postulated that (1) most clinic patients who are non-English speaking and/or setting. lack even a high school education may not follow instructions correctly, and (2) physicians take a more aggressive approach to private patients because they DISCUSSION This study evaluated the quality of care given by believe that they may be more educated, and thus individual physicians to patients with CAD and also more receptive and compliant. Although this may be examined whether different practice settings repre- true, at this medical center such postulated differences CORONARY ARTERY DISEASE/LIPID MANAGEMENT
1419
did not seem to result in differences in management practices or outcomes. The overall results of this study, which demonstrated less than desirable lipid-lowering management in patients with CAD, were similar to those in previous reports. A study by Boekeloo et al1 found that only 53% of hospital inpatient charts had a documented cholesterol level compared with an average of 91% reporting an assessment of blood pressure in patients with CAD. A study by Houston-Miller et al,4 examining lipid management in post-MI patients, demonstrated better lipid control in patients treated by nurses who were given a lipid-lowering protocol than in those whose physicians were not given such a protocol. Last, similar to findings noted by Superko et al,2 we found some variation in cholesterol management between specialties. This study illustrates 2 major points: lipid management quality is independent of the socioeconomic status of the patient, and individual physician behavior is the most determinant factor in delivering quality care. We have shown that preventive services for cardiovascular disorders are dramatically underused and must be improved. The use of protocol-driven lipid-lowering regimens4 and better definition of the physician ultimately responsible for lipid-lowering in the patient with CAD, as well as frequent reviews and feedback on physician performance.12,14,15 physician performance14 –16 will heighten physician awareness and result in better preventive care.
1420 THE AMERICAN JOURNAL OF CARDIOLOGYT
VOL. 81
1. Boekeloo BO, Becker DM, Le Bailly A, Pearson TA. Cholesterol management
in patients hospitalized for coronary heart disease. Am J Prev Med 1988;4:128 – 132. 2. Superko HR, Desmond DA, deSantos VV, Vranizan KM, Farquhar JW. Blood cholesterol treatment attitudes of community physicians: a major problem. Am Heart J 1988;116:849 – 855. 3. Marcelino JJ, Feingold KR. Inadequate treatment with HMG-CoA reductase inhibitors by health care providers. Am J Med 1996;100:605– 609. 4. Houston-Miller N, Thomas RJ, Superko HR, Ghandour G, Taylor B, DeBusk RF. Lipid-lowering therapy in post-MI patients: efficacy of a nurse managed intervention (abstr). Circulation 1991;84(suppl II):SII-328. 5. Friedwald WT, Levy RI, Frederichson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18:499 –502. 6. Wilkinson, L. SYSTAT: The System for Statistics. Evanston, IL: Systat Inc., 1990. 7. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Summary of the Second Report of the National Cholesterol Education Program (NCEP). JAMA 1993;269:3015–3023. 8. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Second Report of the Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel II). Circulation 1994;89:1329 –1445. 9. The Post Coronary Artery Bypass Graft Trial Investigators. The effect of aggressive lowering of low density lipoprotein cholesterol levels and low-dose anticoagulation on obstructive changes in saphenous-vein coronary-artery bypass grafts. N Engl J Med 1997;336:153–162. 10. Safran, DG, Tarlov AR, Rogers WH. Primary care performance in fee-forservice and prepaid health care systems. JAMA 1994;271:1579 –1586. 11. Murray JP, Greenfield S, Kaplan SH, Yano EM. Ambulatory testing for capitation and fee-for-service patients in the same practice setting: relationship to outcomes. Med Care 1992;30:252–261. 12. Boekeloo BO, Becker DM, Levine DM, Belitsos PC, Pearson TA. Strategies for increasing house staff management of cholesterol with inpatients. Am J Prev Med 1990;6:51–59. 13. Riley GF, Potosky AL, Lubitz JD. Stage of cancer as diagnosis for Medicare HMO and fee-for-service enrollees. Am J Public Health 1994;84:1598 –1604. 14. Smith DA, Schall PL. Improved hypertension control using a purveillance system in a neighborhood health center. Med Care 1980;18:774 –776. 15. Schectman JM, Elinsky EG, Bartman BA. Primary care clinician compliance with cholesterol treatment guidelines. J Gen Int Med 1991;6:121–125.
JUNE 15, 1998
MD,
We undertook a study to determine whether there were differences in the quality of lipid management in patients with coronary artery disease (CAD) in 2 different practice settings (which represent different socioeconomic classes), and to determine the level of compliance with the National Cholesterol Education Program guidelines by academic physicians in managing patients with CAD. A retrospective cross-sectional study was performed using a systematic chart review of 270 medical records (131 from the cardiology clinic, 139 from the cardiology private practice) of patients with known CAD at an academic tertiary care center in New York City. The total proportion of patients with CAD having a lipid profile ordered in the clinic and private suite was 43%. Of these people, 22% had a low-density lipoprotein cholesterol (LDL) <100 mg/dl and 54% had an LDL <130 mg/dl (10% and 23% of the total population, respectively). The total proportion of patients taking lipid-lowering medications was 29%. When comparing the quality of treatment between the 2 settings, there were
no statistically significant differences in the percentages of patients who had lipid profiles measured (40% clinic vs 47% private suite, p >0.10), in the percentage of patients with LDL <130 mg/dl (50% clinic vs 57% private suite, p >0.10) or in the weighted percentage of patients taking lipid-lowering medications (29% clinic vs 48% private suite, p 5 0.099). The performances of individual physicians, however, varied widely. The percentages of patients with lipid profiles measured by individual physicians ranged from 0% to 83%, while the percentages of patients on drug treatment by a physician ranged between 10% and 88%. These findings indicate that socioeconomic differences, represented by different practice settings, do not account for differences in the screening for, control of, or use of medications in managing hyperlipidemia. Rather, individual physicians are accountable for differences in lipid management. Q1998 by Excerpta Medica, Inc. (Am J Cardiol 1998;81:1416 –1420)
everal previous studies have documented less than optimal physician compliance with regard to folS lowing the National Cholesterol Education Program
lowering treatment according to the national recommendations and if these patients were receiving a similar quality of care depending on the type of health care setting they visited.
(NCEP) guidelines for lipid management.1–3 These studies, which analyzed the issue of physicians’ attitudes and practices toward secondary prevention of CAD, produced similar results and concluded that no group of physicians reflected an aggressive cholesterol reduction attitude. One investigator found that physicians ordered lipid profiles in only 18% of their patients with CAD,1 whereas another study found that physicians were likely to begin drug treatment for hypercholesterolemia only at total cholesterol levels ranging from 296 to 319 mg/dl.2 Furthermore, they determined that physician practices vary among one another, especially between different specialties and that the management of patients with existing CAD by all specialties is not in accordance with the NCEP guidelines.2 Another trial of patients on monotherapy with HMG coenzyme A reductase inhibitors to reduce cholesterol demonstrated only 33% of patients achieving NCEP guidelines.3 In this study we examined whether this patient population was receiving lipidFrom the Cardiovascular Institute, and the Department of Community Medicine, Mount Sinai School of Medicine, The Mount Sinai Medical Center, New York, New York. Manuscript received September 22, 1997; revised manuscript received and accepted February 2, 1998. Address for reprints: Donald A. Smith, MD, MPH, Mount Sinai School of Medicine, Cardiovascular Institute, PO Box 1014, One Gustave L. Levy Place, New York, New York, 10029.
1416
©1998 by Excerpta Medica, Inc. All rights reserved.
METHODS
Setting: Our study was conducted at both the cardiology clinic and private faculty cardiology suite at the Mount Sinai Medical Center in New York City. This major urban teaching hospital is located on the upper east side of Manhattan and lies adjacent to Spanish Harlem in New York City. The 2 different practice settings that we examined differ mainly in the financial reimbursements that each receives. The private suite receives 40% of its reimbursements from private sources, whereas the clinic receives only 14% of its revenues from private sources (Table I). Similarly, Medicare reimbursements are greater in the private suite (45%) than in the clinic (31%). On the other hand, .50% of the reimbursements that the clinic receives are either from Medicaid or are nonreimbursable visits, whereas only 15% of the private suites’ income is from these sources. Thus, the reimbursement characteristics of each setting depict the different socioeconomic characteristics of the patients who attend them, with a larger proportion of Medicaid patients attending the clinic and more affluent classes attending the private faculty practice. Patients: A cross-sectional study of patients with existing CAD was performed using a chart review 0002-9149/98/$19.00 PII S0002-9149(98)00209-4
TABLE I Source and Percentage of Reimbursement by Setting Office Site (% of funding)
Reimbursement Source
Private Clinic Suite (n 5 131) (n 5 139)
Private funding Private insurance Health Maintenance Organization Self-pay Subtotal, private source Medicare Subtotal, private source 1 Medicare Medicaid Nonreimbursable visits
16 1 1 18 41 59 39 33
(12%) 38 (27%) (1%) 14 (10%) (1%) 4 (3%) (14%) 56 (40%) (31%) 62 (45%) (45%) 118 (85%) (30%) 21 (15%) (25%) 0 (0%)
method. CAD was defined by a history of 1 of 4 conditions: (1) a myocardial infarction, as documented by a physician note in the chart; (2) a coronary artery bypass graft; (3) a percutaneous transluminal coronary angioplasty; or (4) a moderately or severely positive thallium stress test, as documented in the chart by a nuclear cardiologist. In the faculty practice, subjects were selected by reviewing each physician’s charts in an alphabetical manner, looking for the first 15 patients with known CAD. In the clinic, subjects were randomly found by searching through numbered patients charts in ascending order. Patients in both settings were included in the study if they fulfilled the following criteria: (1) had at least 3 outpatient visits to the same physician at the Mount Sinai Hospital between December 1993 and December 1994; and (2) met 1 of the criteria for CAD as mentioned above. In some cases, 15 subjects could not be found for each physician in each setting, and thus the numbers of patients varied by physician and setting. The hospital patient number, gender, and present age of each of the patients meeting these criteria were recorded. The charts of these patients were then reviewed, focusing particularly on 2 factors: (1) if a lipid profile had been ordered within the last 12 months, and (2) if the patient had currently been prescribed a lipid-lowering pharmacologic agent. To ascertain the presence and values of a lipid profile ordered through the hospital laboratory, the hospital patient number was used to reference the hospital laboratory test computer. This was a necessary step as physicians often failed to record the lipid profile order within the charts. Also, many of the charts did not include the most recent lab profiles, so the current lipid values for patients needed to be obtained by the computer. Physicians: Physicians who participated in this study are salaried faculty at the Mount Sinai Hospital and have patients in both the cardiology clinic and the private faculty cardiology suite. In addition to their base salary, the physicians earn a small percentage of their total income generated in the faculty practice. Lipid profiles: The lipid profiles obtained in Mount Sinai Hospital were performed in the following man-
ner: (1) Total cholesterol, high-density lipoprotein (HDL) cholesterol and triglyceride measurements were obtained on the hospital’s Technicon Chem 1 system (Bayer Diagnostics, Tarrytown, New York) according to the manufacturer’s recommendation. (2) Low-density lipoprotein (LDL) cholesterol was calculated using the Friedwald formula.5 Quality control is monitored through the use of bilevel controls run 3 times daily plus periodic surveys conducted by the American College of Pathologists and the New York State Department of Health. Lipid profiles performed within the hospital accounted for most of the tests in both settings. Some lipid profiles of the patients attending the private cardiology suite, however, were performed by outside clinical laboratories. Data analysis: Data were entered into a Lotus 1-2-3 spread sheet, from which simple descriptive statistics were generated. Hypothesis tests were calculated using SYSTAT6 version 5.0 for DOS. Crude comparisons between groups on continuous variables were made using the Student t test, and for categorical variables using the Pearson chi-square and MantelHaenszel chi-square statistics. Linear regression was used for adjusted comparisons of continuous variables across groups, and logistic regression was used for adjusted comparison of categorical variables across groups.
RESULTS
Demographic information: PATIENT PROFILE: We reviewed 270 patient charts from the Mount Sinai Hospital. One hundred thirty-one were obtained from the cardiology clinic and 139 from the private practice suite. All patients were previously diagnosed with CAD as stated in the Methods section. The age (mean 6 SD) of clinic and of the private suite patients was 61 6 10 and 65 6 11 years, respectively (p ,0.003). In the clinic, nearly equal numbers of charts of men and women were reviewed (49%: 51%), whereas more charts of men were reviewed in the private suite (66%:34%). Of the patients who had an LDL measured (n 5 117), clinic patients had a mean LDL of 133 mg/dl and private patients had a mean LDL of 132 mg/dl (p .0.10). Because LDL values were positively skewed, a linear regression analysis using log LDL values revealed the geometric mean LDL cholesterol level to be 1.004 (95% confidence interval 0.88 to 1.14) times as high in the clinic versus the private practice setting, adjusting for age, gender, and individual physician. The power of the model to detect a 5% difference in the LDL cholesterol level by site was 80%. In contrast, women (n 5 55) had a higher mean LDL (145 6 48 mg/dl) than men (n 5 66, 122 6 38 mg/dl), (p ,0.02), a significant difference that persisted when adjusted for differences in age, clinic versus private setting, and the treating physician. A further analysis that showed a higher proportion of women who had an LDL .100 mg/dl (p ,0.05) confirmed this trend. However, the percentage of women with an LDL .100 was no longer statistically CORONARY ARTERY DISEASE/LIPID MANAGEMENT
1417
TABLE II Lipid Management Results by Setting
Lipid profiles in last 12 months LDL cholesterol #130 mg/dl LDL cholesterol #100 mg/dl On lipid-lowering medications Mantel Haenszel chi-square (weighted statistic) LDL cholesterol .130 mg/dl on lipid-lowering medications
Overall (n 5 270)
Clinic (n 5 131)
117 63 26 77
52 26 11 29
(43%) (54%) (22%) (29%)
26 (50%)
significant after applying the Mantel-Haenszel stratification analysis to control for the physician (p 5 0.09). Thus, although overall there are more women than men likely to have an LDL .100, this apparently happens because physicians who are less likely to achieve an LDL #100 mg/dl are treating more of the women. PHYSICIAN PROFILE: The physicians participating in our study are attending physicians at the Mount Sinai Medical Center who see patients in both the clinic and the private practice suite. A total of 11 physicians participated in our study. Ten physicians are cardiologists and 1 is an endocrinologist/lipids specialist with additional expertise in public health. Their years in practice range from 1 to 15 (average 6.3). PHYSICIAN PRACTICE PATTERNS: The overall physician performance was compared with the recommended 1993 NCEP guidelines for managing hypercholesterolemia in patients with CAD. The following parameters were examined: (1) presence of a lipid profile within the past 12 months; (2) percentage of patients with LDL ,100 mg/dl; (3) percentage of patients with LDL ,130 mg/dl; (4) percentage of patients taking a lipid-lowering medication; and (5) percentage of patients with LDL .130 mg/dl on lipidlowering medication. Of all 270 patients, 117 (43%) had a lipid profile in the preceding 12 months. Of these patients, 26 (22%) had an LDL ,100 mg/dl and 63 (54%) had an LDL ,130 mg/dl (10% and 23% of the total population, respectively). The proportion of all patients on a lipidlowering medication was 29% (n 5 77). Among the 153 patients without a lipid profile, 26 were nevertheless receiving lipid-lowering medication. Of patients with an LDL .130 mg/dl, exactly 50% received drug treatment (Table II). When examining the frequency of ordering lipid profiles by individual physicians, considerable variability is evident between physicians, but apparently less variation within each physician’s own practice patterns (Figure 1). For example, physician A ordered lipid profiles in approximately 70% of his patients (of all patients seen), whereas physician J ordered lipid profiles only in about 10% of his (data not shown). Apparently, some physicians ordered more lipid profiles in the clinic, as seen in physicians A, D, G, and K, whereas some ordered more in the private suite (Figure 1). The trend within each physician’s ordering 1418 THE AMERICAN JOURNAL OF CARDIOLOGYT
VOL. 81
(40%) (50%) (21%) (22%)
10 (40%)
Private Suite (n 5 139) 65 37 15 48
(47%) (57%) (23%) (35%)
16 (59%)
p Value .0.10 .0.10 .0.10 ,0.05 0.099 .0.10
of lipid profiles was relatively constant between the 2 settings for each physician. Note that the lipids specialist, physician K, ordered lipid profiles more frequently than any of the cardiologists (83% of his total patients had a lipid profile, data not shown). When analyzing the rates of lipid-lowering medication prescriptions by physicians, a similar pattern was noted (Figure 2). Once again, the physicians differed widely between one another, prescribing medications to between 10% and 50% of their patients, and in general the cardiologists prescribed less medications than the lipids specialist (prescribed to 88% of patients, data not shown). Note that some physicians prescribe to more patients in the clinic and others prescribe to more in the private suite, as was the pattern with the ordering of lipid profiles. Although, there are occasional physicians for whom the difference between settings are great (physicians B, D, and H), in general they are consistent between the 2 settings (Figure 2). It is also interesting to note that physicians A, B, and C, who ordered more lipid profiles, tended to prescribe more lipid-lowering medications. COMPARISON OF CARE PROVIDED TO PATIENTS IN CLINIC AND IN THE PRIVATE SUITE: There was no statis-
tically significant difference in the proportions of patients who had lipid profiles performed (40% clinic vs 47% private suite, p .0.10), in the percentage of patients with lipid profiles who had an LDL #130 mg/dl (50% clinic vs 57% private suite, p .0.10), and in the percentage with an LDL #100 mg/dl (21% clinic vs 23% private suite, p .0.10) (Table II). The proportion of patients taking lipid-lowering medications did differ substantially in the 2 settings, however, with 22% of clinic patients taking medications compared with 35% of private practice patients taking medications (p ,0.05) (Table II). To see if this was caused by variability in numbers of patients seen by specific physicians in the 2 settings, the original 2 3 2 analysis was subdivided into 11 smaller 2 3 2 analyses (1 for each physician). The association between drug presence or absence in the clinic versus the private suite was tested by individual physician and then combined using the Mantel-Haenszel chi-square statistic. This summary statistic, which weighs each physician equally, correcting for any differences in the numbers of patients seen by each physician in each setting was not significant (chi-square 5 2.723, p 5 0.099), demonstrating no difference in the use of JUNE 15, 1998
senting the patients’ socioeconomic status further influenced the quality of care. Our results demonstrated 3 important findings. First, there were no significant differences in the management of hyperlipidemia in patients with CAD between a hospital clinic and private practice setting. Second, physicians who participated in this study did not adequately comply with the 1993 NCEP guidelines. In fact, ,50% of the study patients had a lipid profile performed within 12 months of treatment, of which 78% had an LDL cholesterol level FIGURE 1. Percentage of patients with lipid profiles by individual physician. Physicians A above the recommended level of through J: cardiologists. Physician K: endocrinologist. The numbers below the horizontal 100 mg/dl for secondary prevenaxis are the numbers of patient charts reviewed in each setting. Filled bars above each tion.7–9 Although at the time of letter represent the percentages of clinic patients by physician; open bars represent the this study a target LDL level of percentages of private patients by physician. The last column on the right represents the percentage of patients in the clinic (filled) and in the private suite (open). comb. 5 com,130 mg/dl was recommended bined. for secondary prevention, we found 46% of the study patients to have values above this goal. Last, our study revealed that lipid management varies more widely among individual physicians than between different practice settings. One possible reason for only partial compliance of specialists with the NCEP is that lipid management is left to the prerogative of the primary care physician. This is evident in our study because 17% of patients without a documented lipid profile by their cardiologist were receiving lipidlowering therapy. Previous studies have shown FIGURE 2. Percentage of patients on lipid-lowering therapy by physician. Physicians A differences in quality of care bethrough J: cardiologists. Physician K: endocrinologist. The numbers below the horizontal tween different care areas, namely axis are the numbers of patient charts reviewed in each setting. Filled bars above each letter represent the percentages of clinic patients by physician; open bars represent the Medicaid/Medicare clinics and percentages of private patients by physician. The last column on the right represents the fee-for-service private pracpercentage of patients in the clinic (filled) and in the private suite (open). comb. 5 com10 –13 The flaws in these studtices. bined. ies, which were circumvented in this study, included the use of subjective measurements,11 indrugs in the clinic or private suite. In effect, we found equality of services offered to different patients,12 and that physicians who typically prescribe lipid-lowering the use of different physicians in different settings medication less frequently (in either setting) ac- which confounded a simple setting comparison.13 Similarities in the management of lipids in 2 difcounted for more of the patients seen in the clinic setting, resulting in an apparent, but deceptive, overall ferent care areas with 2 different patient populations in finding of less medications prescribed in the clinic this study were surprising. We postulated that (1) most clinic patients who are non-English speaking and/or setting. lack even a high school education may not follow instructions correctly, and (2) physicians take a more aggressive approach to private patients because they DISCUSSION This study evaluated the quality of care given by believe that they may be more educated, and thus individual physicians to patients with CAD and also more receptive and compliant. Although this may be examined whether different practice settings repre- true, at this medical center such postulated differences CORONARY ARTERY DISEASE/LIPID MANAGEMENT
1419
did not seem to result in differences in management practices or outcomes. The overall results of this study, which demonstrated less than desirable lipid-lowering management in patients with CAD, were similar to those in previous reports. A study by Boekeloo et al1 found that only 53% of hospital inpatient charts had a documented cholesterol level compared with an average of 91% reporting an assessment of blood pressure in patients with CAD. A study by Houston-Miller et al,4 examining lipid management in post-MI patients, demonstrated better lipid control in patients treated by nurses who were given a lipid-lowering protocol than in those whose physicians were not given such a protocol. Last, similar to findings noted by Superko et al,2 we found some variation in cholesterol management between specialties. This study illustrates 2 major points: lipid management quality is independent of the socioeconomic status of the patient, and individual physician behavior is the most determinant factor in delivering quality care. We have shown that preventive services for cardiovascular disorders are dramatically underused and must be improved. The use of protocol-driven lipid-lowering regimens4 and better definition of the physician ultimately responsible for lipid-lowering in the patient with CAD, as well as frequent reviews and feedback on physician performance.12,14,15 physician performance14 –16 will heighten physician awareness and result in better preventive care.
1420 THE AMERICAN JOURNAL OF CARDIOLOGYT
VOL. 81
1. Boekeloo BO, Becker DM, Le Bailly A, Pearson TA. Cholesterol management
in patients hospitalized for coronary heart disease. Am J Prev Med 1988;4:128 – 132. 2. Superko HR, Desmond DA, deSantos VV, Vranizan KM, Farquhar JW. Blood cholesterol treatment attitudes of community physicians: a major problem. Am Heart J 1988;116:849 – 855. 3. Marcelino JJ, Feingold KR. Inadequate treatment with HMG-CoA reductase inhibitors by health care providers. Am J Med 1996;100:605– 609. 4. Houston-Miller N, Thomas RJ, Superko HR, Ghandour G, Taylor B, DeBusk RF. Lipid-lowering therapy in post-MI patients: efficacy of a nurse managed intervention (abstr). Circulation 1991;84(suppl II):SII-328. 5. Friedwald WT, Levy RI, Frederichson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18:499 –502. 6. Wilkinson, L. SYSTAT: The System for Statistics. Evanston, IL: Systat Inc., 1990. 7. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Summary of the Second Report of the National Cholesterol Education Program (NCEP). JAMA 1993;269:3015–3023. 8. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Second Report of the Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel II). Circulation 1994;89:1329 –1445. 9. The Post Coronary Artery Bypass Graft Trial Investigators. The effect of aggressive lowering of low density lipoprotein cholesterol levels and low-dose anticoagulation on obstructive changes in saphenous-vein coronary-artery bypass grafts. N Engl J Med 1997;336:153–162. 10. Safran, DG, Tarlov AR, Rogers WH. Primary care performance in fee-forservice and prepaid health care systems. JAMA 1994;271:1579 –1586. 11. Murray JP, Greenfield S, Kaplan SH, Yano EM. Ambulatory testing for capitation and fee-for-service patients in the same practice setting: relationship to outcomes. Med Care 1992;30:252–261. 12. Boekeloo BO, Becker DM, Levine DM, Belitsos PC, Pearson TA. Strategies for increasing house staff management of cholesterol with inpatients. Am J Prev Med 1990;6:51–59. 13. Riley GF, Potosky AL, Lubitz JD. Stage of cancer as diagnosis for Medicare HMO and fee-for-service enrollees. Am J Public Health 1994;84:1598 –1604. 14. Smith DA, Schall PL. Improved hypertension control using a purveillance system in a neighborhood health center. Med Care 1980;18:774 –776. 15. Schectman JM, Elinsky EG, Bartman BA. Primary care clinician compliance with cholesterol treatment guidelines. J Gen Int Med 1991;6:121–125.
JUNE 15, 1998