Lipid screening to prevent coronary artery disease: a quantitative evaluation of evolving guidelines [CMAJ

Canadian Medical Association Journal

Home

Back issues
Supplements
Selected series

eLetters
About this journal
Info for authors

Lipid screening to prevent coronary artery disease: a quantitative evaluation of evolving guidelines
authors
Steven A. Grover,* Marc Dorais,* Gilles Paradis,† J. George Fodor,‡ Jiri J. Frohlich,§ Ruth McPherson,¶ Louis Coupal,* Hanna Zowall*
CMAJ 2000;163(10):1263-9 [PDF]

Abstract

Background: There is strong evidence to support the treatment of abnormal blood lipid levels among people with cardiovascular disease. Primary prevention is problematic because many individuals with lipid abnormalities may never actually develop cardiovascular disease. We evaluated the 1998 Canadian lipid guidelines to determine whether they accurately identify high-risk adults for primary prevention.

Methods: Using data from the Lipid Research Clinics and receiver operating characteristic (ROC) curves, we compared the diagnostic performance of the 1998 lipid guidelines when risk factors for coronary artery disease (CAD) were counted versus calculating risk using Framingham risk equations. We also compared the diagnostic accuracy of the 1998 guidelines with guidelines previously published by the National Cholesterol Education Program in the United States and the 1988 Canadian Consensus Conference on Cholesterol and then used Canadian Heart Health Survey data to forecast lipid screening and treatment rates for the Canadian population.

Results: The Framingham risk equations were more accurate than counting risk factors for predicting CAD risk (areas under the ROC curves, 0.83 [standard deviation (SD) 0.02] v. 0.77 [SD 0.03], p < 0.05). Risk counting was a particularly poor method for predicting risk for women. The 1998 Canadian guidelines identified high-risk individuals more accurately than the earlier guidelines, but the increased accuracy was largely due to a lower false-positive rate or a higher true-negative rate (i.e., increased test specificity). Using the 1998 lipid guidelines we estimate that 5.9 million Canadians currently free of cardiovascular disease would be eligible for lipid screening and 322 705 (5.5%) would require therapy.

Interpretation: Calculating risk using risk equations is a more accurate method to identify people at high risk for CAD than counting the number of risk factors present, especially for women, and the 1998 Canadian lipid screening guidelines are significantly better at identifying high-risk patients than the 1988 guidelines. Many of our findings were incorporated into the new 2000 guidelines.

Contents
• • • • • •

Despite progress in prevention and treatment, cardiovascular disease remains a leading cause of death and disability among Canadians.¹ Epidemiologic studies have identified important cardiovascular risk factors including hypertension, diabetes mellitus, a sedentary lifestyle, obesity, cigarette smoking, elevated low-density lipoprotein (LDL) cholesterol and depressed high-density lipoprotein (HDL) cholesterol.^2,3,4 Many of these risk factors, including blood lipids, are modifiable and amenable to treatment.^{5,6,7,8,9,10,11} This has resulted in increasing interest in the benefits of lipid therapy, as well as rising health care expenditures associated with lipid-lowering medication.^12,13,14

A decision to treat abnormal blood lipid levels in those who have already developed cardiovascular disease is a relatively simple one given the clinical trial results to date.^7,8,11 Primary prevention is more problematic, however, because many individuals with lipid abnormalities may never actually develop cardiovascular disease, or they may develop it only after many years.^5,6,9,10 Also, physicians may not be particularly accurate at estimating cardiovascular risk on the basis of an intuitive evaluation of an individual patient.¹⁵ Treatment guidelines have therefore been developed to help physicians target high-risk patients for primary prevention therapy.

In 1988 the Canadian Consensus Conference on Cholesterol (CCCC) published national guidelines for screening and treating hyperlipidemia.¹⁶ Similar guidelines have been published in the United States^17,18 and Europe.¹⁹ Although there are substantial differences between these guidelines, their common purpose is to identify individuals who are most likely to benefit from the treatment of lipid abnormalities.

Health Canada recently completed an initiative to update lipid screening and treatment guidelines for Canada. In 1998 the Working Group on Hypercholesterolemia and Other Dyslipidemias published the preliminary results of these deliberations²⁰ emphasizing that the evolving guidelines were a working document open to amendments, revisions and adjustments based on new research and a careful evaluation of the guidelines themselves. We present a quantitative evaluation of the 1998 screening guidelines and compare them with guidelines published in the United States by the National Cholesterol Education Program (NCEP I and NCEP II)^17,18 and in Canada by the CCCC.¹⁶ Our analyses of the accuracy of the guidelines identify potential areas for improvement, many of which have been incorporated into the new 2000 guidelines.³⁰

Contents
• • • • • •

Methods

We used data provided by the Lipid Research Clinics (LRC) Prevalence and Follow-up Studies^21,22,23 to evaluate the diagnostic performance of the 1998 screening guidelines. We also used the Canadian Heart Health Survey²⁴ (CHHS) data to estimate the prevalence of cardiovascular risk factors among the adult population in Canada and the number of individuals who would require lipid therapy.

LRC Program Prevalence Study

The LRC Prevalence Study was conducted from 1972 to 1976 in 10 North American clinics to determine the prevalence of lipid abnormalities.^21,22,23 The focus of our analysis is a 15% random sample of participants who were selected to provide a representative cross-section of the population. We excluded individuals with cardiovascular disease, subjects not eligible for screening (i.e., men younger than 40 or older than 70 years of age and women younger than 50 or older than 70), those suspected of having coronary artery disease (CAD) because they were taking digitalis or antiarrhythmics, individuals who were pregnant or were taking lipid medications and those for whom we did not have all laboratory test results. After these exclusions, data for 2218 (52%) of the randomly selected participants were analysed.

Participants were followed for an average of 12.2 years, and specific causes of death were verified using death certificates. Because nonfatal coronary events were not documented in the LRC study, we used death due to CAD as a proxy for coronary risk. Framingham Study data²⁵ demonstrated that these 2 outcomes share the same risk factors. For our analyses "coronary deaths" included those classified as definite or suspected CAD-related deaths according to the LRC protocol.^21,22,23

The 1998 guidelines recommend that target values for blood lipid levels be based on an assessment of the overall risk of a cardiovascular event using 1 of 2 strategies.²⁰ The first requires a 10-year risk calculation based on multivariate equations published by the Framingham Heart Study,^3,25 also known as the multivariate risk approach. The second strategy, referred to as the risk-counting approach, estimates CAD risk on the basis of the number of risk factors present.²⁰ For example, using the 1998 screening guidelines, patients with 4 or more risk factors or with a calculated 10-year risk of 40% or more are considered at very high risk for a cardiac event (Table 1); among such very high-risk individuals treatment should be initiated if their LDL-cholesterol level is greater than or equal to 3.5 mmol/L or the total cholesterol:HDL-cholesterol ratio is greater than 5 (Table 2).

Diagnostic performance of the 1988 guidelines

We used receiver operating characteristic (ROC) curves²⁶ to compare the true-positive rate (sensitivity) and false-positive rate (1 – test specificity) of the various guidelines when only LRC subjects classified above a specific decision threshold (low, moderate, high or very high risk) were considered to be at increased risk for CAD-related death. For example, a fixed number of CAD-related deaths occurred over 12.2 years, a proportion of which occurred among very high-risk individuals. The higher this proportion, the better the true-positive rate of the guidelines if only very high-risk individuals are eligible for treatment. However, some of those classified as very high-risk who died of noncardiac-related causes would be needlessly concerned about their blood lipid levels (i.e., the false-positive rate). This exercise can be repeated at different thresholds (e.g., where only those at high or very high risk are treated). An ROC curve plots the true-positive rate on the y axis and the corresponding false-positive rate on the x axis for each treatment threshold. The area under the resulting fitted curve represents the diagnostic performance or discriminating ability of the guidelines. We compared the diagnostic performance of the 4 different guidelines by comparing the areas under their ROC curves.²⁶

Canadian Heart Health Surveys data

The Canadian Heart Health Surveys (CHHS) represent the main findings of 9 provincial surveys conducted between 1986 and 1990.²⁴ We analyzed the data for men between 40 and 70 years of age and women between 50 and 70 years of age with no symptomatic CAD or stroke (n = 4174). The risk factors menopause, family history of premature CAD and left ventricular hypertrophy were not included in our analyses because the data were not collected in the CHHS. Glucose levels were also absent from the CHHS data; for our purposes this variable was replaced with diabetic status.

Each adult sampled in the CHHS was assigned a weight for the number of Canadians he or she represented.²⁴ These weights, based on the 1992 Canadian census, were used when forecasting the results of implementing the 1998 screening guidelines nationwide.

Contents
• • • • • •

Results

Risk profiles of the LRC follow-up cohort and the CHHS participants

The 1998 Canadian lipid screening guidelines focus primary prevention screening efforts on 40- to 70-year-old men and 50- to 70-year-old women. A comparison of the cardiovascular disease risk profiles of the LRC and CHHS participants is presented in Table 3. Although the mean ages of the men and women were slightly lower in the LRC population, the distribution of risk factors in the 2 samples was remarkably consistent.

Diagnostic accuracy of counting risks versus calculating multivariate risk

The accuracy of the 2 proposed risk assessment strategies was compared using ROC curve analyses on 40- to 70-year-old men and 50- to 70-year-old women in the LRC cohort. The risk-counting strategy demonstrated an overall discriminating ability of 0.77 (standard deviation [SD] 0.03), which was significantly lower (p < 0.05) than the 0.83 (SD 0.02) obtained using the multivariate risk-assessment approach. Further analyses demonstrated that this was largely because of the poor performance of the risk-counting strategy for predicting CAD among women compared with men (discriminating ability 0.59 [SD 0.12] for women v. 0.81 [SD 0.03] for men). Results of the multivariate risk-assessment approach were similar for men and women (discriminating ability 0.83 and 0.82, respectively).

Comparing guidelines

We compared the diagnostic accuracy of the 1998 lipid screening guidelines with the CCCC¹⁶ guidelines and the American NCEP I and NCEP II recommendations for predicting CAD in the LRC cohort.^17,18 Although these earlier guidelines also targeted adults 30 years of age and older, we restricted our analyses to the age groups in the 1998 guidelines (i.e., 40- to 70-year-old men and 50- to 70-year-old women).

Overall, the discriminating ability of the 1998 screening guidelines risk-counting strategy was significantly better than that of the CCCC (0.77 [SD 0.03] v. 0.66 [SD 0.04], p < 0.05) (Fig. 1). The 1998 guidelines were also marginally better than the NCEP I (0.69 [SD 0.04], p = 0.06) and the NCEP II (0.70 [SD 0.04], p = 0.15).

However, on closer inspection the better discrimination of the 1998 guidelines was related to a higher true-negative rate (or improved test specificity) at the expense of a lower true-positive rate (or reduced test sensitivity). For example, earlier guidelines classified approximately 19%–26% of LRC adults as high risk and suitable for treatment, whereas the 1998 guidelines would target only 10.5% of adults for treatment (Table 4). Raising the requirements for being classified as suitable for treatment increases the true-negative rate of the 1998 screening guidelines to 90% compared with 75%–82% for older guidelines.

Projected results of screening Canadians

Using the 1998 guidelines and the CHHS data, we projected that approximately 5.9 million Canadians free of known cardiovascular disease would be eligible for lipid screening (Table 5). This number represents approximately 3.7 million men between 40 and 70 years of age and 2.2 million women between 50 and 70. Given that risk counting is more likely to be adopted by busy physicians than multivariate risk assessment, we projected our results using the risk-counting approach. Less than 1%, or 18 646 men, would be identified as very high risk for CAD with 4 or more risk factors, whereas 61% (2 237 684 men) would be classified as low risk with either 1 or no risk factors. The results for Canadian women were similar.

The number of Canadians who would require treatment was also forecasted using the threshold lipid values (Table 2) for initiating treatment (see Table 6). Overall, 322 705 adults or 5.5% of the Canadian screened would require therapy. Among the 26 237 Canadians with 4 or more risk factors 11 904 (45.4%) would undergo treatment; 16 246 (0.5%) of the the 3.6 million Canadians classified as low risk would require treatment.

Contents
• • • • • •

Interpretation

The 1998 lipid screening guidelines discriminate more accurately between people at high risk for CAD and those at low risk for CAD than the old CCCC guidelines.^16,20 This increased accuracy is largely due to a lower false-positive rate. However, the associated higher false-negative rate means more people, who would eventually die from cardiac-related causes, would be falsely assured that they did not require treatment. This problem could be overcome in part by recalibrating the treatment guidelines to more aggressively treat those at the lower levels of risk for CAD. For instance, lipid target levels for each risk level could be lowered so that more individuals would be eligible for treatment. Unfortunately, this would increase the false-positive rate as well.

Because the LRC data included only fatal coronary events, it was necessary to substitute death due to CAD for total CAD risk when evaluating the diagnostic performance of the 1998 guidelines. Nonetheless, this approach can be used to compare the relative accuracy of different lipid screening guidelines, given that the risk factors for death due to CAD have been shown to also predict CAD events.²⁵ We also note that some risk factors specified in the guidelines were not available in the CHHS data for evaluation. However, these risk factors, including family history and left ventricular hypertrophy were also not identified as independent risk factors in the recent Framingham analyses.²⁵ Finally, we emphasize that we focused on primary prevention screening only and did not consider the current guidelines for secondary prevention lipid treatment.

The 1998 screening guidelines also demonstrate marginally superior discriminating ability when compared with their American counterparts.^17,18 It is not readily apparent why this is so, given that they all consider lipid parameters and non-lipid risk factors. However, it has been shown that the ratio of total cholesterol to HDL-cholesterol level is the best lipid discriminator for increased risk for CAD;²⁷ the 1998 Canadian guidelines incorporate this important lipid risk factor, whereas the American guidelines focus primarily on LDL-cholesterol levels, with a more minor role for HDL cholesterol.¹⁸

If the 1988 guidelines are applied to the CHHS data and extrapolated to the Canadian population, approximately 5% of Canadian adults eligible for screening would ultimately be targeted for some form of lipid therapy. These results are consistent with a similar analysis by Haq and colleagues²⁸ in which British Health Survey data and Framingham equations were used to estimate the proportion of the British adult population that would be eligible for lipid therapy. Among men with no cardiovascular disease between 40 and 69 years of age, they estimated 0.6% would have a 10-year risk of CAD of over 45%, 6.8% would have a risk of over 30%, and 32% would have a risk of 15% or more. In our analysis of Canadian men we estimate that 0.5% would have a risk of 40% or more, 6.3% would have a risk of 20%–39%, and 32.2% would have a risk of 10%–19%. For women between 50 and 70 years of age, the British study estimated that 0% would have a risk of 45% or more, 0.8% would have a risk of 30% or more and 17.9% would have a risk of 15% or more. The estimates for Canadian women are: 0.3% with a 10-year risk of over 40% or more, 5.7% would have a risk of 20%–39%, and 31.6% would have a risk of 10%–19%.

These consistent results indicate that the overall 10-year risk for CAD for most people with hyperlipidemia in our society is below 10%. Accordingly, treatment decisions based on short-term risk estimates will result in a substantial portion of the population not receiving therapy for blood lipid levels that are significantly abnormal. These data underscore the need to develop primary prevention strategies on the basis of not only short-term risk but also the potential long-term benefits of treatment over a lifetime.²⁹

Previous analyses of the CCCC guidelines indicated that over 50% of the Canadian population would be targeted for lipid therapy if only blood lipids levels were considered.¹⁶ However, when applying the 1998 guidelines approximately 5% of the population would require lipid therapy. Which is more correct? Clinical trial data alone cannot dictate who should be treated. The long-term benefits, the associated risks and the net cost of treatment must also be considered when these decisions are made. Also, if we are to improve long-term compliance, patients' preferences must not be ignored.

Clearly, the 1998 lipid screening guidelines represent an ongoing effort to formulate a clinically reasonable, scientifically sound and affordable strategy to treat hyperlipidemia in Canadians. This study was conducted to provide a systematic evaluation of the guidelines, which was needed to advance the development of this primary prevention policy. The updated 2000 guidelines for the management and treatment of dyslipidemia³⁰ have incorporated many of our observations including the importance of accurately assessing the risk for a coronary event.

Competing interests: None declared for Drs. Dorais, Paradis, Coupal and Zowall. Drs. Grover, Fodor, Frohlich and McPherson have received speaker fees and travel assistance from various pharmaceutical companies. Dr. Grover has also received investigator-initiated research grants from various companies; Dr. Fodor, industrial research grants for work in this area; and Dr. Frohlich, consultant fees and honoraria from companies dealing with lipid-lowering agents.

Contributors: All authors contributed to the preparation, revision and approval of the final manuscript.

Acknowledgements: Dr. Grover is a senior research scholar (Chercheur-Boursier) supported by the Fonds de la recherche en santé du Québec. This investigator-initiated study was supported in part by a non-restricted grant from Fournier Canada; at no time did employees of Fournier Canada provide input into the study analysis, results or conclusions.

Contents
• • • • • •

From *the Centre for the Analysis of Cost-Effective Care and the Divisions of General Internal Medicine and Clinical Epidemiology, The Montreal General Hospital, and the Departments of Medicine and Epidemiology and Biostatistics, McGill University; †the Division of Preventive Medicine, McGill University Health Centre and the Department of Epidemiology and Biostatistics, McGill University, Montreal, Que.; ‡the Prevention and Rehabilitation Centre and ¶the Division of Cardiology, University of Ottawa Heart Institute, Ottawa, Ont.; and §the Healthy Heart Program, St. Paul's Hospital, Vancouver, BC.

This article has been peer reviewed.

Reprint requests to: Dr. Steven A. Grover, The Montreal General Hospital, Division of Clinical Epidemiology, 1650 Cedar Ave., Montreal QC H3G 1A4

References

    1.    Heart and Stroke Foundation of Canada. Heart disease and stroke in Canada. In: Reeder B, Chockalingam A, Dagenais G, MacLean D, Nair C, Petrasovits A, et al., editors. Ottawa: The Foundation; 1997.

    2.    Abbott RD, McGee D. The probability of developing certain cardiovascular disease in eight years at specified values of some characteristics. In: Kannel WB, Wolf PA, Garrison RJ, editors. The Framingham Study: an epidemiological investigation of cardiovascular disease. Washington: US Department of Health, Education and Welfare; 1987. Cat no. NIH 87-2284.

    3.    Castelli WP, Garrison RJ, Wilson PW, Abbott RD, Kalousdian S, Kannel WB. Incidence of coronary heart disease and lipoprotein cholesterol levels. The Framingham Study. JAMA 1986;256(20):2835-8. [MEDLINE]

    4.    Grover SA, Palmer C, Coupal L. Serum lipid screening to identify high-risk individuals for coronary death: the results of the Lipid Research Clinics Prevalence Cohort. Arch Intern Med 1994;154(6):679-84. [MEDLINE]

    5.    The Lipid Research Clinics Coronary Primary Prevention Trial results. I. Reduction in incidence of coronary heart disease. JAMA 1984;251(3):351-64. [MEDLINE]

    6.    Frick MH, Elo O, Haapa K, Heinonen OP, Heinsalmi P, Helo P, et al. Helsinki Heart Study: primary-prevention trial with gemfibrozil in middle-aged men with dyslipidemia. Safety of treatment, changes in risk factors, and incidence of coronary heart disease. N Engl J Med 1987;317(20):1237-45. [MEDLINE]

    7.    Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet 1994;344:1383-9. [MEDLINE]

    8.    Sacks FM, Pfeffer MA, Moye LA, Rouleau JL, Rutherford JD, Cole TG, et al. The effect of pravastatin on coronary events after myocardial infarction in patients with average cholesterol levels. N Engl J Med 1996;335:1001-9. [MEDLINE]

    9.    Shepherd J, Cobbe SM, Ford I, Isles CG, Lorimer AR, MacFarlane PW, et al. Prevention of coronary heart disease with pravastatin in men with hypercholesterolemia. N Engl J Med 1995;333:1301-7. [MEDLINE]

    10.    Downs JR, Clearfield M, Weis S, Whitney E, Shapiro DR, Beere PA, et al. Primary prevention of acute coronary events with lovastatin in men and women with average cholesterol levels: results of AFCAPS/TexCAPS. JAMA 1998;279:1615-22. [MEDLINE]

    11.    The Long-Term Intervention with Pravastatin in Ischaemic Disease (LIPID) Study Group. Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and a broad range of initial cholesterol levels. N Engl J Med 1998;339(19):1349-57. [MEDLINE]

    12.    Pedersen TR, Kjekshus J, Berg K, Olsson G, Wilhelmsen L, Wedel H, et al. Cholesterol lowering and the use of healthcare resources. Results of the Scandinavian Simvastatin Survival Study. Circulation 1996;93(10):1796-802. [MEDLINE]

    13.    Johannesson M, Jonsson B, Kjekshus J, Olsson AG, Pedersen TR, Wedel H. Cost effectiveness of simvastatin treatment to lower cholesterol levels in patients with coronary heart disease. Scandinavian Simvastatin Survival Study Group. N Engl J Med 1997;336:332-6. [MEDLINE]

    14.    Grover SA, Coupal L, Paquet S, Zowall H. Cost-effectiveness of 3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors in the secondary prevention of cardiovascular disease: forecasting the incremental benefits of preventing coronary and cerebrovascular events. Arch Intern Med 1999;159(6):593-600. [MEDLINE]

    15.    Grover SA, Lowensteyn I, Esrey K, Steinert Y, Joseph L, Abrahamowicz M. How accurately do Canadian physicians assess the coronary risk of their patients? The preliminary results of the Coronary Health Assessment Study (CHAS). BMJ 1995;310:975-8. [MEDLINE]

    16.    Canadian Consensus Conference on Cholesterol: final report. The Canadian Consensus Conference on the prevention of heart and vascular disease by altering serum cholesterol and lipoprotein risk factors. CMAJ 1988;139(11 Suppl):1-8. [MEDLINE]

    17.    Report of the National Cholesterol Education Program Expert Panel on detection, evaluation, and treatment of high blood cholesterol in adults. The Expert Panel. Arch Intern Med 1988;148(1):36-69. [MEDLINE]

    18.    Summary on the second report of the National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation, and treatment of high blood cholesterol in adults. JAMA 1993;269(23):3015-23. [MEDLINE]

    19.    Prevention of coronary heart disease in clinical practice. Recommendations of the second joint task force of European and other societies on coronary prevention. Eur Heart J 1998;19(10):1434-503. [MEDLINE]

    20.    Frohlich J, Fodor G, McPherson R, Genest J, Langner N, for the Dyslipidemia Working Group of Health Canada. Rationale for and outline of the recommendations of the Working Group on Hypercholesterolemia and Other Dyslipidemias: interim report. Can J Cardiol 1998;14(Suppl A):17A-21A.

    21.    Central Patient Registry and Coordinating Center for the Lipid Research Clinics. Reference manual for the Lipid Research Clinics Prevalence Study. vols 1 and 2. Chapel Hill (NC): Department of Biostatistics, University of North Carolina; 1974.

    22.    Plasma lipid distributions in selected North American populations: the Lipid Research Clinics Prevalence Study. The Lipid Research Clinics Program Epidemiology Committee. Circulation 1979;60:427-39. [MEDLINE]

    23.    Heiss G, Tamir I, Davis CE, Tyroler HA, Rifkind BM, Schonfeld G, et al. Lipoprotein-cholesterol distributions in selected North American populations: the Lipid Research Clinics Program Prevalence Study. Circulation 1980;61:302-15. [MEDLINE]

    24.    MacLean DR, Petrasovits A, Nargundkar M, Connelly PW, MacLeod E, Edwards A, at al. Canadian heart health surveys: a profile of cardiovascular risk. Survey methods and data analysis. Canadian Heart Health Surveys Research Group. CMAJ 1992;146(11):1969-74. [MEDLINE]

    25.    Anderson KM, Wilson PWF, Odell PM, Kannel WB. An updated coronary risk profile. A statement for health professionals. Circulation 1991;83:356-62. [MEDLINE]

    26.    Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148(3):839-43. [MEDLINE]

    27.    Grover SA, Coupal L, Hu XP. Identifying adults at increased risk of coronary disease: How well do the current cholesterol guidelines work? JAMA 1995;274:801-6. [MEDLINE]

    28.    Haq I, Ramsay LE, Pickin DM, Yeo WW, Jackson PR, Payne JN. Lipid-lowering for prevention of coronary heart disease: what policy now? Clin Sci 1996;91(4):399-413. [MEDLINE]

    29.    Grover SA. Gambling with cardiovascular disease: picking the winners and losers [editorial]. Lancet 1999;353:254-5. [MEDLINE]

    30.    JG Fodor, JJ Frohlich, JJG Genest, Jr, PR McPherson, for the Working Group on Hypercholesterolemia and Other Dyslipidemias.Recommendations for the management and treatment of dyslipidemia. CMAJ 2000;162:1441-7. [MEDLINE]