How Effective Is Drug Treatment Of J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from Hypercholesterolemia? A Guided Tour Of The Major Clinical Trials For The Primary Care Physician

Kevin Grumbacb, M.D.

Abstrtlet: Bllcllgt'OfllUl: Drug trea1ment of hypercholesterolemia remains con1rOversial. Central to the debate are the results of the major pJ.acebo-con1rOlled clinical tria1s of pharmacologic trea1ment of hypercholesterolemia. Methods: Conventions and principles of clinical epidemiology are used to review the four major clinical tria1s of drug treatment of hypercholesterolemia. The review trans1ates the results of these large, epidemiologically oriented experiments into terms that are applicable to managing patients at the individual level. Results: Clofibrate is an ineffective trea1ment. Primary prevention with gemftbrizol or cholestyramine requires treating approximately 50 middle-aged men for 10 years to avert one adverse outcome. Secondary prevention with nJacln for men with a prior myocardial infarction requires treatment of 10 to 15 patients for 10 years to prevent one adverse outoome. Conclusions: While drug treabnent of hypercholesterolemia in middle-aged men can prevent deadl and morbidity, the magnitude of the effectiveness is modest. Because a critical factor inftuenclng the magnitude of benefit is the underlying risk of adverse events in the population under trea1ment, physicians should 1arget interventions to populations that may benefit the most. In populations for whom the magnitude of effectiveness is likely to be modest, physicians should exercise clinical judgment when deciding what degree

of benefit justifies treatment in individual cases. 0 Am Board Fam Pract 1991; 4:437-45.) copyright.

Drug treatment of hypercholesterolemia remains dispute, or is the debate over issues of interpreta­ controversial. "While many assume that the Na­ tion and clinical judgment? Is it possible to quan­ tional Cholesterol Education Program Panel tify objectively the benefits and risks of drug treatment guidelines 1 define the standard of care treatment? Are the results of controlled research for hypercholesterolemia, reports in both the trials applicable to the real world of clinical professional and lay literature have challenged practice? http://www.jabfm.org/ these guidelines as being overly aggressive in Insights into the answers to these questions can recommending pharmacologic interventions.2-8 result from examining the major clinical trials that Adding to the confusion, advocates of either ag­ constitute much of the empirical basis for gressive or conservative practices frequently in­ hypercholesterolemia treatment guidelines. Part voke the results of the same clinical trials to de­ of the challenge in examining these trials, how­ fend widely disparate points of view. ever, is translating the results of large, epidemio­ on 27 September 2021 by guest. Protected \Vhat accounts for the disagreement over treat­ logically oriented experiments into terms appli­ ment of hypercholesterolemia? Do the results of cable to caring for patients at the individual level. the major clinical trials conflict? Are the facts in This article discusses the principles and conven­ tions of clinical epidemiology that facilitate this translation and then uses these principles to Submitted, revised, 21 June 1991. evaluate the major trials of drug treatment of From the Department of Family and Community Medicine and the Institute for Health Policy Studies, University of Califor - hypercholesterolemia. nia, San Francisco. Dr. Grumbach was a Pew Health Policy Fellow at the time this article was written. Address reprint re­ Clinical 'I\iaIs in Perspec1ive quests to Kevin Grumbach, M.D., San Francisco General Hospi­ tal, Bldg. 1, Room 202,1001 Potrero Avenue, San Francisco, CA Large, population-based studies, such as the Fra­ 94110. mingham project, have established that persons

Drug Treatment of Hypercholesterolemia 437 J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from Table l. Relative and Absolute Risk Reduction: Hypotbetic:a1 Treatment Jiffectiveness Examples. A prev~ntive intervention is effective if it is able to Percent reduce the occurrence of adverse clinical out­ Adverse Relative Risk Absolute Risk comes. One way to express effectiveness is as the Example Events' Reductiont Reduction't proportion of adverse events averted by treat­ 1. Placebo 40 75 30 ment-the relative risk reduction. Alternatively, Drug 10 one can formulate effectiveness as the actual num­ 2. Placebo 4 75 3 ber of adverse events prevented by treatment, or Drug I the absolute risk reduction. I I 3. Placebo 14 21 3 Drug 11 Table 1 illustrates the different information 'per 100 subjects. communicated when effectiveness is expressed as t Relative risk reduction = [(Rateplacebo - RateRx)lRateplacebo] either relative risk reduction or absolute risk re­ X 100%. duction. In example 1, 40 of 100 untreated pa­ tAbsolute risk reduction =Rateplacebo - RateRx. tients experienced an adverse outcome compared with hypercholesterolemia have a significantly in­ with only 10 of 100 treated patients. Treatment creased probability of developing cardiovascular thus prevented 75 percent ([40-10]/40) of the disease and of dying at a younger age than those events, representing the relative risk reduction. In with lower cholesterol levels.9•lo* In addition, absolute terms, treatment of 100 patients pre­ many short-term studies have shown that phar­ vented 30 untoward outcomes-the absolute risk macologic treatment can reduce cholesterol levels reduction. In the second example, only 4 of 100 to a normal range in persons with elevated levels. untreated patients had adverse events while 1 of Why, in the face of these two bodies of evidence, 100 treated patients had such an event. The rela­ is it necessary to perform long-term trials costing tive risk reduction is 75 percent, the same as in millions of dollars to evaluate drug treatment? If example 1; however, the absolute risk reduction is persons with lower cholesterol levels are at lower only 3 events averted per 100 patients treated copyright. risk of disease, and if drugs reduce cholesterol compared with 30 per 100 in example 1. levels, is it not reasonable to assume that drug Examples 1 and 2 highlight one of the basic treatment will reduce the incidence of disease? challenges facing primary care physicians at­ There are a number of reasons to exercise cau­ tempting to interpret the results of clinical trials. tion in making this assumption. Having normal Investigators tend to emphasize effectiveness in levels of cholesterol may not be the same as hav­ terms of relative risk reduction, while clinicians

ing levels normalized by drug treatment. Al­ tend to be more interested in knowing how many http://www.jabfm.org/ though associated with a higher risk of cardiovas­ of the patients they treat will actually benefit (the cular disease, hypercholesterolemia may not be absolute risk reduction).2.11 As shown in Table 1, the actual cause of disease. For example, choles­ treatment of the same relative effectiveness may terol simply might be a marker for some uniden­ produce a tenfold difference in absolute effective­ tified dietary or physiologic factor that is the true ness, depending on the underlying rate of adverse instigator of atherosclerosis, and reducing serum events in the untreated population. on 27 September 2021 by guest. Protected cholesterol levels with drug treatment might not Example 3 in Table 1 shows how a treatment of correct the underlying process. In addition, drug less relative effectiveness than the treatment in treatment might produce side effects that partly example 2 could still yield an absolute risk reduc­ or completely offset the gains achieved by risk tion of 3 per 100 simply by treating a population factor reduction. For these reasons, among with a higher underlying risk of adverse outcomes others, it is important to perform trials that care­ (14 per 100 as opposed to 4 per 100). fully measure the actual clinical outcomes of drug Relative risk reduction is primarily a function intervention. of the intrinsic efficacy of the intervention (how well the drug itself works), compliance with the drug (how faithfully the patient takes the drug), 'Although there is general agreement that hypercholesterole­ and the proportion of adverse outcomes caused by mia is a risk factor for cardiovascular disease and overall mortality in middle-aged men. whether hypercholesterolemia raises the the risk factor in the first place (the attributable risk in older men and in women of any age is not entirely clear. 4.10 risk). Attributable risk limits the ultimate effec-

438 JABFP Nov.-Dec. 1991 Vol. 4 No.6 tiveness of treatment for a multifactorial disease, the overall quality of life fared among treated and J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from such as coronary artery disease, because even the control subjects, not exclusively about cardiovas­ best cholesterol-reducing medication will not en­ cular morbidity; tirely avert progression of coronary disease that is Considering all clinically meaningful outcomes due to, for example, continued smoking. is important, because in several of the trials re­ Absolute effectiveness is a product of both the viewed here, reductions in cardiovascular deaths relative risk reduction and the underlying rate of are offset by increases in noncardiovascular adverse events. As example 2 in Table 1 shows, deaths. Clinical trials of risk factor interventions, even a treatment that is highly efficacious in rela­ often requiring thousands of patients and many tive terms may yield modest benefit if the under­ years of treatment to show measurable benefit, lying incidence rate is low. In other words, even may detect toxicities that were not apparent in the the best medication is of limited benefit if few more limited studies required for initial drug people are going to experience bad events even licensing. without treatment. Absolute risk reduction best In presenting the results of the major clinical reflects the primary care physician's vantage point trials, this review focuses on the outcomes of in relating the number of patients benefiting to (1) death from any cause, (2) cardiovascular the number actually started on treatment. morbidity (nonfatal myocardial infarctions), and Laupacis and colleaguesll have proposed (3) combined all-cause mortality and cardio­ an even simpler formulation of absolute risk re­ vascular morbidity. Also highlighted are reports duction, which they refer to as the number of showing significant differences in noncardio­ patients needed to treat to avert one adverse out­ vascular morbidity between treated and control come. The number needed to treat is simply the groups. reciprocal of the absolute risk reduction. In exam­ ple 1 of Table 1, the absolute risk reduction of Review of the Major Clinic:al 'Irials 30 events prevented for every 100 patients Selection CrlterItI treated may be reformulated as needing to treat In selecting published trials for this review, only copyright. approximately 3 (100/30) patients to avert one those studies adhering to a relatively rigorous event. definition of the classic randomized, placebo­ For each of the studies in this review, the fig­ controlled clinical trial-the "gold standard" of ures for relative risk reduction, absolute risk re­ experimental design-were chosen. The selection duction, and number needed to be treated are criteria were as follows: provided. 1. The study included a pharmacologic inter­

vention for hypercholesterolemia, rather http://www.jabfm.org/ CII"kIIl Outcomes than dietary treatment alone. If effectiveness defines a treatment's success at 2. The study randomized assignment of subjects preventing adverse outcomes, what are the to treatment and control groups. clinical outcomes that merit consideration? 3. The study compared active treatment with Obviously, investigators conducting hypercho­ placebo treatment. 4. The primary outcomes were death and clini­ lesterolemia treatment trials are particularly in­ on 27 September 2021 by guest. Protected terested in measuring cardiovascular outcomes, cally apparent disease, rather than less-overt such as myocardial infarctions and strokes. Pub­ outcomes, such as the status of coronary ath­ lished reports typically highlight reductions in erosclerosis viewed by angiogram. 12 cardiovascular deaths, nonfatal cardiovascular events, and total cardiovascular events. lIyperdIolaterolBmlll TrlIIIs Primary care physicians, not to mention pa­ Tables 2 and 3 summarize the four major tients, may have a slightly broader perspective on hypercholesterolemia treatment trialS. 13- 16 The meaningful clinical outcomes. The ultimate ques­ World Health Organization (WHO), Lipid Re­ tion-Did patients receiving active treatment live search Clinics (LRC), and Helsinki Heart studies longer?-tends to be more relevant than knowing all examined primary prevention, i.e., treat­ only about deaths from a particular cause. Simi­ ment of subjects with no overt manifestations larly, many of us are interested in knowing how of cardiovascular disease. The Coronary Drug

Drug Treatment of Hypercholesterolemia 439 J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from 'bbIe Z. ~ 'niall-Study ChancterIItIca.*

Study Subjects Initial Cholesterol Level Drug

WH013 Men only Average .. 6.45 mmollL (250 mg/dL) Clofibrate 30-59 years old (top 33% of sampled population) No overt cardiovascular disease LRCI4 Men only Average =7.55 mmollL (292 mg/dL) Cholestyramine 33-59 years old (top 5% of sampled population) No overt cardiovascular disease Average LDL =5.53 mmollL (216 mg/dL) No hypertension Helsinki IS Men only Average .. 7.00 mmollL (270 mg/dL) Gemfibrizol 40-55 years old (top 20% of sampled population) No overt cardiovascular disease Average LDL .. 4.86 mmollL (190 mg/dL) CDpI6 Men only Average .. 6.45 mmollL (250 mg/dL) Clofibrate or niacin 30-60 years old (not selected for high cholesterol) 1 or more prior myocardial infarctions

Project (COP) studied secondary prevention in of their cholesterol level, although the average subjects with a history of myocardial infarction. cholesterol level was relatively high (6.45 mmollL All of the trials examined middle-aged men [250 mgldL]). exclusively. Drug treatments included clofibrate, niacin, The three primary prevention trials selected cholestyramine, and gemfibriwl. No major study men with elevated total cholesterol levels using of clinical outcomes has been reported with criteria ranging from the upper 33 percent of lovastatin or oat bran. Subjects both on active screened values in the WHO study to the upper drug and placebo treatment received dietary 5 percent of values in the LRC study. The counseling. In all studies, cholesterol levels de­

COP trial of secondary prevention selected men clined to a significantly greater degree in subjects copyright. with a prior myocardial infarction irrespective on active treatment.

Table 3. HyperdIoIestaoIe 'MIll-Clinical Outannea.

Relative Risk Absolut Number Study Years of Reduction Risk Needed To Follow-up Outcomes Incidence* (%) Reduction* Treatt

Placebo Drug http://www.jabfm.org/

WH013 5 Death 19 25 -24 -6 -167 NonfiltalMI 31 23 26 8 125 Death or nonfatal MI 50 48 NS LRCI4 7.4 Death 37 36 NS NonfiltalMI 118 102 14 16 63 Death or nonfatal MI 149 135 9 14 71

Helsinki IS 5 Death 21 22 NS on 27 September 2021 by guest. Protected NonfiltalMI 35 22 37 13 77 Death or nonfatal MI 56 44 20 12 83 CDpI6 Niacin group 6.2 Death 254 244 NS NonfiltalMI 138 102 26 36 28 Death or nonfatal MI 392 346 12 46 22 15* Death 582 520 11 62 16 Qofibrate group 6.2 Death 254 255 NS NonfiltalMI 138 131 NS 15:1: Death 582 578 NS MI =myocardial infilrction. NS .. no statistically significant difference between placebo and drug groups. *Incidence and absolute risk reductions expressed as events per thousand subjects. tNumber of subjects needed to be treated to avoid one event. *15-year follow-up study conducted after completion of treatment phase (only overall mortality measured).

440 JABFP Nov.-Dec. 1991 Vol. 4 No.6 J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from As indicated in Table 3, not one of the studies nonfatal outcomes, type of ongoing medical showed a significant reduction in overall mor­ and pharmacologic treatment, or health behav­ tality among subjects on active drugs during the ior for the years between the conclusion of study period. In the WHO study, overall mor­ the formal trial and the follow-up ascertain­ tality was significantly higher among subjects ment of vital status. It is difficult therefore to taking clofibrate because of an excess of non­ be certain whether the diverging death rates rep­ cardiovascular deaths. In the LRC (cholestyra­ resent delayed direct effects of niacin treat­ mine) and Helsinki (gemfibrizol) studies, small ment, indirect effects of the initially lower nonfa­ decreases in cardiovascular mortality were coun­ tal myocardial infarction rate translating into terbalanced by small increases in noncardiovascu­ lower mortality with time, or the effects of un­ lar mortality. specified treatments and behaviors in the inter­ In terms of nonfatal outcomes, all four studies vening years. consistently showed a benefit of treatment in pre­ To allow direct comparison of the effectiveness venting nonfatal myocardial infarctions. The of treatment among the hypercholesterolemia number needed to treat to avert one nonfatal trials, the results were extrapolated to a uniform myocardial infarction ranged from 28 patients for 10-year duration of treatment for all the studies 6.2 years in the CDP-niacin group to 125 patients (using the methods of Laupacis, et al.ll). These for 5 years in the WHO study. Treatment had no 10-year "numbers needed to treat" to prevent impact on the incidence of stroke and peripheral either death or a nonfatal cardiovascular event vascular disease. are shown in Figure 1. These projections should In addition to its uniqueness as a trial of sec­ be regarded as estimates of the general mag­ ondary prevention, the CDP study had a follow­ nitude of benefit of treatment in the various up component that distinguishes it from the other studies. studies. 17 Several years after the CDP trial was

formally concluded, investigators attempted to Side Bffocts of1'retllrlunlt copyright. determine whether there were any long-term Table 4 presents many of the symptoms or clinical or delayed effects of treatment on overall mor­ conditions reported to occur more frequently in tality. At the 15-year follow-up, overall mortal­ subjects on active drug in the hypercholesterole­ ity was significantly lower among subjects mia trials. The format follows that of the morbid­ who had received niacin while they were en­ ity and mortality tables, but with headings of rolled in the trial. Unfortunately, the CDP fol­ relative risk and absolute risk (rather than risk low-up study included no information about reduction). Number needed to treat refers to the http://www.jabfm.org/

Table 4. Trealmmt Side Effects. Number Relative Absolute Needed to Study Rx Group Outcomes Incidence* Risk Risk* Treatf Placebo Drug WH013 Clofibrate Cholecystectomy 5 11 2.2 6 167 on 27 September 2021 by guest. Protected CDp16 Clofibrate Cholelithiasis I3 30 2.3 17 59 Impotence or decreased 100 141 1.4 41 24 libido Niacin Gout 43 64 1.5 21 48 Flushing 43 920 21.4 877 1 Skin abnormalities 158 263 1.7 105 10 Abdominal pain 143 214 1.5 71 14 Helsinki 15 Gemfibrizol "Moderate to severe upper 70 113 1.6 43 23 gastrointestinal symptoms" (1st year) LRC14 Cholestyramine "Moderate to severe 430 680 1.6 250 4 gastrointestinal side effects" (1st year) *Incidence and absolute risk expressed as events per thousand subjects. tNumber of subjects needed to be treated to produce one side effect.

Drug Treatment of Hypercholesterolemia 441 J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from Number Needed To T... t subjects in the Helsinki study. The magnitude of eo.-----~~~------_, this difference in underlying event rate far out­ eo weighs "differences in factors influencing relative risk reduction (e.g., the different efficacies of par­ 40 ticular drugs) in determining the absolute effec­ 30 tiveness in these studies. On the other hand, the lack of any relative risk reduction with clofibrate 20 in both the WHO and CDP studies indicates that

10 clofibrate is an inherently ineffective drug (e.g., INEFFECTIVE with no benefit or with risk exceeding benefit) no o LIIC CDI'-Cloflllr." WHO matter what the underlying event rate. The difference in the magnitude of effective­ Figure 1. Number of patients needed to be treated for 10 ness between the niacin-CDP study of secondary years to avert 1 adverse event. (COP-Niacin = Coronary prevention and the three primary prevention Drug Project, niacin group; Helsinki = Helsinki Heart trial hypercholesterolemia trials shows one of the [gemftbrizol); LRC = Upid Research Clinics trial [cbolestynunine); CDP-Ciotlbrate = Coronary Drug great ironies of preventive medicine: secondary Project, dotlbrate group; WHO = World Health prevention is often more effective than primary Orgaoimtion trial [dotlbrate).) prevention in terms of the magnitude of absolute risk reduction because it targets a subpopulation that has declared itself to be at particularly high number of patients receiving treatment that is risk of future events. Saying that secondary pre­ expected to result in one adverse side effect occur­ vention is more effective than primary prevention ring. Side effects range from mild and reversible, may sound heretical. Indeed, preventing a second but extremely common, occurrences (e.g., flush­ myocardial infarction may be a lesser accomplish­

ing with niacin) to more severe but less frequent ment than preventing the first infarction. My copyright. conditions (e.g., cholelithiasis with clofibrate). point is not to discredit the laudable goal of pri­ mary prevention; nevertheless, primary preven­ Discussion tion invariably increases the number needed to Targeting S"bpojnlllltkms lit JBgbest RisII treat to prevent an event because many persons What accounts for the superior absolute effec­ are treated who were never destined to experience tiveness of treatment with niacin in the CDP cardiovascular disease. study? As Table 3 indicates, this result is not due In the best of all possible worlds, clinicians http://www.jabfm.org/ to a greater relative risk reduction in the niacin­ could predict exactly which patients will develop CDP study. For the outcome "death or cardiovas­ cardiovascular disease and limit risk factor inter­ cular morbidity," the 12 percent relative risk re­ ventions to those individuals. Because currently duction in the niacin-CDP group was comparable available measures to define risk, such as blood with the 9 percent relative risk reduction in the pressure or cholesterol levels, are often at best LRC trial; both were considerably less than the 20 rough predictors of adverse clinical outcomes, percent relative risk reduction in the Helsinki clinicians face treating risk factors in many pa­ on 27 September 2021 by guest. Protected study. The key factor explaining the difference in tients who would have done equally well (or bet­ absolute risk reduction is the much higher under­ ter) without treatment. As these clinical trials in­ lying rate of adverse events in the CDP study. The dicate, treating risk factors in populations with underlying event rate in the CDP placebo group low underlying rates of adverse events leads to (392 of 1000) was nearly three times the rate in large numbers of patients needing to be treated to the LRC placebo group (149 of 1000) and six avert a single event. This will be the case with the times the rate in the Helsinki placebo group (56 best of medications, i.e., one producing a high of 1000). Clearly, subjects in the CDP trial were relative risk reduction. For primary care physi­ different. They had already suffered a myocardial cians, then, selecting the population of patients to infarction, placing them at much higher risk of target for treatment becomes as important as or death or subsequent infarction, even though their even more important than selecting the particular average cholesterol level was less than that of drug for treatment.

442 JABFP Nov.-Dec. 1991 Vol. 4 No.6 The recommendations of the National Choles­ 'While some persons are eager to "do everything J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from terol Education Project (NCEP)1 attempt to tar­ possible" to prevent disease and willingly accept get subpopulations at particular risk by including the tradeoffs oflong-term medication, others are risk factors other than just hypercholesterolemia less favorably inclined to medical intervention. As into many of the treatment algorithms for hyper­ Table 4 shows, drug side effects are relatively cholesterolemia. It should be noted, however, common with many of the agents. Patients clearly that subjects in the hypercholesterolemia trials will differ in their attitudes about tolerating some had many of the additional risk factors featured immediate and potentially self-limited discomfort in the NCEP guidelines: all were men, and some for the chance of averting some major mishap in smoked, had family histories of premature car­ the future. There is evidence that the act of de­ diovascular disease, and so on. The numbers tecting hypertension and labeling an individual as needed to treat listed in Figure 1 may be reason­ hypertensive may result in deteriorating psycho­ able estimates of the effectiveness of treating the logical well-being and increased work absentee­ types of populations targeted by the NCEP ism. 19,20 Patients also have different views about guidelines. the "medicalization" of their lifestyles entailed by the use of medications and the frequent visits and Bow Effective SbofIltl tl1'retItmtmI Be, laboratory tests that accompany treatment. Brett,2 in a commentary on hypercholesterole­ Medicalization is particularly important when the mia, describes drug treatment of cardiovascular population considered for treatment is asymp­ risk factors as "an enterprise in probabilities that tomatic. As Brett and others note,2,21 drug side incorporates scientific data, the patient's values, effects and issues of medicalization make taking a and the patient's attitude toward medical inter­ medication very different from giving up fatty ventions while asymptomatic. "P 679 The results of foods or cigarettes. the major clinical trials of drug treatment of These considerations should be part of every hypercholesterolemia provide some of the best physician-patient deliberation, no matter what copyright. scientific data about treatment effectiveness. In the diagnosis or treatment. 22 But they take on the case of severe hypertension, a risk factor for increasing importance when an intervention of­ which studies have shown that the number fers a 1-in-50 chance of benefit rather than a needed to treat to avert a death or morbid cardio­ 1-in-5 chance. Among both physicians and pa­ vascular event is only 5 patients for 1.5 years, 18 the tients, disagreements about the overall balance of scientific data may be so compelling as to elimi­ risk and benefit are inevitable when the magni­ nate all controversy about the merits of treat­ tude of treatment effectiveness is modest. ment. But in instances when effectiveness appears Examining magnitudes of effectiveness can also http://www.jabfm.org/ more modest, questions of values assume greater help physicians to prioritize their clinical activi­ prominence. A critical factor in treatment deci­ ties. How can a physician make best use of the sions becomes the primary care physician's judg­ 10 to 15 minutes spent with a patient during an ment about what level of effectiveness justifies office visit? In general, clinical trials suggest that treatment in individual cases. treating mild to moderate hypertension is more

How should a physician and a patient decide effective than treating hypercholesterolemia in on 27 September 2021 by guest. Protected whether effectiveness in the range of a 1-in-50 asymptomatic individualsP-2s Facing a patient chance of benefiting from treatment of hyper­ with both hypertension and hypercholesterole­ cholesterolemia is worth it? The choice to inter­ mia, a physician might want to emphasize re­ vene may be straightforward when a preventive ducing blood pressure-for example, by prioritiz­ intervention, such as a vaccine, is cheap, relatively ing antihypertensive medication if the patient nontoxic, and minimally disruptive to patients' experiences difficulty tolerating both anti­ lives. Unfortunately, long-term drug treatment of hypertensive and cholesterol-lowering medi­ risk factors is considerably more complicated cations. Counseling a patient to quit smoking when evaluating the tradeoffs of intervention. probably has greater benefit than drug treatment Patients and physicians must weigh the possible of either mild to moderate hypertension or compromises in quality of life that may accom­ hypercholesterolemia.26 Even so, more than pany the potential benefits of taking medication. one-half of the smokers who visited a physician

Drug Treannent of Hypercholesterolemia 443 in the previous year state their physician never and disadvantages of preventive treatment for J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from advised them to quit smoking.27 To the extent individual patients and when setting clinical that preoccupation with interventions for hyper­ priorities. cholesterolemia distracts physicians from coun­ seling patients to discontinue smoking, physician resources are inefficiently used. References It is instructive to compare American consensus 1. Report of the National Cholesterol Education Pro­ reports with those of other countries. Partly on gram Expert Panel on Detection, Evaluation, and the basis of the same clinical trials reviewed Treatment of High Blood Cholesterol in Adults. The Expert Panel. Arch Intern Med 1988; 148: herein, one British panel has rejected mass cho­ 28 36-69. lesterol screening and other British panels have 2. Brett AS. Treating hypercholesterolemia. How been much more cautious in advising drug treat­ should practicing physicians interpret the pub­ ment ofhypercholesterolemia.29,3o A Scottish au­ lished data for patients? N Engl J Med 1989; thority has suggested that drug treatment of 321:676-80. hypercholesterolemia be reserved for secondary 3. Zweig S.Should cholesterol-lowering drugs be used 3 routinely to treat moderate hypercholesterolemia prevention. ! Some Canadian guidelines for drug in patients with serum cholesterol levels of 6.20 to treatment of hypercholesterolemia have also been 6.85 mmollL (240 to 265 mgldL). An opposing view. less vigorous than the guidelines of the American J Fam Pract 1988; 26:670-5. NCEp'32 Physicians should keep in mind the 4. Garber AM, Sox HC Jr, Littenberg B. Screening broader cultural and economic values that influ­ symptomatic adults for cardiac risk factors: the serum cholesterol level. Ann Intern Med 1989; ence the thresholds of effectiveness considered 110:622-39. reasonable for adopting interventions on a wide 5. Garber AM. Where to draw the line against choles­ scale into clinical practice. 33 terol. Ann Intern Med 1989; 111:625-7. 6. Rahimtoola SH. Cholesterol and coronary heart dis­ ease: a perspective.JAMA 1985; 253:2094-5. Conclusion 7. Palumbo PJ. Cholesterol lowering for all: a closer copyright. Whether as a member of an expert panel develop­ look.JAMA 1989; 262:91-2. ing practice guidelines, as a participant in a peer­ 8. Moore 1]. Heart failure. New York: Random review committee for a teaching program or House, 1989. group practice, or as a practitioner developing 9. Dawber TR. The Framingham Study: the epidemi­ individualized patient care plans, physicians must ology of atherosclerotic disease. Cambridge MA: rely on the results of clinical research to make Harvard University Press, 1980. 10. Anderson KM, Castelli WP, Levy D. Cholesterol well-informed decisions about the appropriate­

and mortality. 30 years of follow-up from the Fra­ http://www.jabfm.org/ ness of clinical interventions. The following mingham Study.JAMA 1987; 257:2176-80. principles incorporated into this review of 11. Laupacis A, Sackett DL, Roberts RS. An assessment hypercholesterolemia are equally applicable to of clinically useful measures of the consequences of the analysis of other clinical topics: treatment. N EnglJ Med 1988; 318:1728-33. 12. Blankenhorn DH, Nessim SA, Johnson RL, Sanmarco ME, Azen Sp, Cashin-Hemphill L. Bene­ 1. Focus on the absolute as well as relative risk ficial effects of combined colestipol-niacin therapy

reduction. on coronary atherosclerosis and coronary venous by­ on 27 September 2021 by guest. Protected 2. Scrutinize the inclusiveness and clinical rele­ pass grafts.JAMA 1987; 257:3233-40. vance of the outcomes measured. 13. Oliver MF, Heady JA, Morris IN, Cooper J. A co­ operative trial in the primary prevention of ischae­ 3. Recognize that the underlying probability mic heart disease using clofibrate. Br Heart J 1978; of adverse events in the population studied 40: 1069-1118. exerts a critical influence on the magni­ 14. The Lipid Research Clinics Coronary Primary tude of absolute risk reduction; use this Prevention Trial results. I. Reduction in inci­ information to target treatment to sub­ dence of coronary heart disease. JAMA 1984; populations of patients most likely to benefit 251:351-64. 15. Frick MH, Elo 0, Haapa K, Heinonen OP, from treatment. Heinsalmi P, Helo P, et al. Helsinki Heart Study: pri­ 4. Use estimates of the general magnitude of mary-prevention trial with gemfibrizol in middle­ effectiveness when weighing the advantages aged men with dyslipidemia. Safety of treatment,

444 JABFP Nov.-Dec.I991 Vol. 4 No. 6 changes in risk factors, and incidence of coronary results from the European Working Party on High J Am Board Fam Pract: first published as 10.3122/jabfm.4.6.437 on 1 November 1991. Downloaded from heart disease. N EnglJ Med 1987; 317:1237-45. Blood Pressure in the Elderly Trial. Lancet 1985; 16. Clofibrate and niacin in coronary heart disease. 1:1349-54. ]AMA 1975; 231:360-81. 25. The Australian therapeutic trial in mild hyperten­ 17. Canner PL, Berge KG, Wenger NK, Stamler ], sion. Report by the Management Committee. Lan­ Friedman L, Prineas R], et a1. Fifteen year mortality cet 1980; 1:1261-7. in Coronary Drug Project patients: long-term bene­ 26. Cummings SR, Rubin SM, Oster G. The cost-effec­ fit with niacin.] Am ColI Cardio11986; 8:1245-55. tiveness of counseling smokers to quit.]AMA 1989; 18. Veterans Administration Cooperative Study Group 261:75-9. on Antihypertensive Agents. Effects of treatment on 27. Anda RF, Remington PL, Sienko DG, Davis RM. morbidity in hypertension: results in patients with Are physicians advising smokers to quit? The diastolic blood pressures averaging 115 through patient's perspective.]AMA 1987; 257:1916-9. 129 mmHg.]AMA 1967; 202:1028-34. 28. Smith WC, Kenicer MB, Davis AM, Evans AE, 19. Macdonald LA, Sackett DL, Haynes RB, Taylor Yarnell]. Blood cholesterol: is population screening DW. Labelling in hypertension: a review of the warranted in the UK? Lancet 1989; 1:372-3. behavioural and psychological consequences. ] 29. Leitch D. Who should have their cholesterol Chronic Dis 1984; 37:933-42. concentration measured? What experts in the 20. Haynes RB, Sackett DL, Taylor DW, Gibson ES, United Kingdom suggest. Br Med] 1989; 298: Johnson AL. Increased absenteeism from work after 1615-6. the detection and labeling of hypertensive patients. 30. The British Cardiac Society Working Group on N Engl] Med 1978; 299:741-4. Coronary Prevention: conclusions and recommen­ 21. Rose G. Strategy of prevention: lessons from cardio­ dations. Br Heart] 1987; 57:188-9. vascular disease. Br Med] 1981; 282:1847-51. 31. Oliver MF. Sounding board. Risks of correcting the 22. Eddy DM. Comparing benefits and harms: the bal­ risks of coronary disease and stroke with drugs. N ance sheet.]AMA 1990; 263:2493, 2498, 2501 passim. Engl] Med 1982; 306:297-8. 23. Veterans Administration Cooperative Study Group 32. Naylor CD, Basinski A, Frank Jw. Rachlis MM. on Antihypertensive Agents. Effect of treatment Asymptomatic hypercholesterolemia: clinical review. on morbidity in hypertension: results in patients The Toronto Working Group on Cholesterol Pol­ with diastolic blood pressure averaging 90 through icy.] Clin Epidemioll990; 43:1021-99.

114 mmHg.]AMA 1970; 213:1143-52. 33. Himmelstein DU, Woolhandler S. Free care, copyright. 24. Amery A, Birkenhager W, Brixko P, Bulpitt C, Clem­ cholestyramine, and health policy. N Engl] Med ent D, Deruyttere M, et al. Mortality and morbidity 1984; 311:1511-4. http://www.jabfm.org/ on 27 September 2021 by guest. Protected

Drug Treatment of Hypercholesterolemia 445