Should people at low risk of cardiovascular disease take a statin?

We want to thank Drs. Davis and Dietrich for raising important concerns about our analysis of the effect of statins in people at low risk of cardiovascular disease.

Based on CTT data published in 2012, we showed that statins do not reduce overall mortality in people whose 10-year risk of a major vascular event is less than 20%. Drs. Davis and Dietrich are correct that patients treated with rosuvastatin in the JUPITER trial experienced a significant reduction in all-cause mortality compared to those in the control group. However, the data from this study were included in the CTT meta-analysis of 2012, so singling out the all-cause mortality findings from this study alone would undermine the purpose of the meta-analysis.

That said, there are other issues with the JUPITER trial that merit consideration. At the time the study was prematurely stopped, there was absolutely no difference between the two groups in the number of deaths due to cardiovascular causes—MI and stroke—with 12 such deaths in each group. (This can be calculated by subtracting “Nonfatal myocardial infarction” from “Any myocardial infarction” and doing the same for nonfatal and any stroke from data in Table 3 of the JUPITER trial report in NEJM. ) The entire difference in all-cause mortality in the JUPITER trial was, inexplicably, due to differences in non-cardiovascular mortality.

Further, the JUPITER trial reported more than twice the rate of myocardial infarction in control group patients, but the mortality rates for MI in the two groups are highly anomalous. Again from Table 3 of the NEJM article, patients treated with rosuvastatin had a post-MI mortality rate of 29% (9 fatal MIs/31 total MIs), which is lower than the 40-50% mortality rate that would be expected. But the MI mortality rate in the control group defies credulity: 6 fatal MIs out of a total of 68 MIs, for a mortality rate of 8.8%. This strongly suggests that there was a difference in ascertainment of MI between the two groups. The unrealistically low mortality rate for myocardial infarction in the control group calls into question the accuracy of the myocardial infarction rates. These questions were raised in an article published in the Archives of Internal Medicine in 2010, but there was no response from JUPITER trialists.

We believe that Drs. Davis and Dietrich have conflated “serious adverse events” with “major adverse cardiovascular events.” Serious adverse events— deaths from all causes, hospital admissions, prolongations of admission, cancer, or permanent disability—are measured in all studies, but unfortunately reported in few of the statin studies. The largest of the studies to report the incidence of serious adverse events, JUPITER, found no difference between the active treatment and control groups. The 2013 Cochrane review of statins for primary prevention reports no difference between the incidence of serious adverse events (including cardiovascular events) noted in meta-analyses of all studies that reported serious adverse event rates.

There is not a uniform definition of “MACE” or major adverse cardiovascular events. In our analysis, we chose to limit the definition to “hard events”—cardiovascular death, myocardial infarction, and stroke. These events have a major impact on people’s lives, should have minimal ambiguity in adjudication, and are not vulnerable to geographical variation or bias due to unblinding of treatment allocation (LDL levels are lower in patients in the treatment arms of the studies).

Our analysis in BMJ was published 3 weeks before the release of the updated U.S. Guidelines for cholesterol-lowering. We had no advanced knowledge of the forthcoming recommendations. We based our comments on the 2012 CTT meta-analysis, an associated editorial, and the 2013 Cochrane Review of statins for primary prevention. We had no idea that the updated guidelines would set the threshold for recommending statin therapy at 7.5% 10-year risk of major vascular event guidelines.

And finally, Drs. Davis and Dietrich are mistaken that we discussed people with 10 year risk below 5% “extensively.” Their confusion is understandable because the relevant publications present both 5-year and 10-year percentage risk data. Frankly, BMJ editors asked if we could avoid this potential confusion by presenting either 5-year or 10-year baseline risk data. In our opinion, doing so would have compromised the integrity of the data too much, and we presented the data as they were reported in the original articles. We apologize for the confusion.

We thank the Cochrane review authors for their thoughtful comments (excerpted in bold below), which provide the opportunity to clarify unresolved issues about the benefits and harms of statins in low risk people.

The timing of this Analysis predated by three weeks the publication of the 2013 American College of Cardiology/American Heart Association cholesterol treatment guidelines (November 12, 2013), but statements were made about “proposed standards” without full knowledge of these guidelines.

The “proposed 2013 standards” to which we referred in our Analysis[1] were not the as-yet-unpublished 2013 American College of Cardiology/American Heart Association cholesterol treatment guidelines,[2] but rather the 2013 update of the Cochrane review titled “Statins for the primary prevention of cardiovascular disease,”[3] which had incorporated the findings and recommendations of the 2012 CTT meta-analysis.[4]

The 2012 CTT meta-analysis reported that statin therapy provides a significant reduction in major vascular events in people “with 5-year risk of major vascular events lower than 10%.” It concluded that the then-current major guidelines – ATP-III in the U.S., the European Society of Cardiology task force, and the National Institute of Clinical Excellence in the U.K. – “might need to be reconsidered.”

The 2013 Cochrane Review stated: “….in light of new evidence derived from the CTT Collaboration on primary prevention, there is a need to update existing cost-effective analysis.”

We based our comments about “proposed standards” on these calls to update existing recommendations. Although we had no advanced knowledge of the contents of the forthcoming ACC/AHA cholesterol treatment guidelines, we anticipated that these findings and recommendations would be influential.

Abramson et al. state: "Under the proposed 2013 standards, however, no level of risk would preclude statin therapy…

The 2012 CTT meta-analysis published in the Lancet concluded: “The present report shows that statins are indeed both effective and safe for people with 5-year risk of major vascular events lower than 10% and, therefore, suggests that these guidelines might need to be reconsidered.” No lower limit of benefit is suggested in the CTT meta-analysis.

The 2013 Cochrane Review then stated, “Our previous conclusion urging caution in the use of statins in people at low risk of cardiovascular events is no longer tenable in light of the CTT Collaboration findings.” Again, a lower limit of risk for the benefit of statins was not specified.

…raising the question whether all people over the age of 50 should be treated.” Neither the Cochrane review nor the ACC/AHA cholesterol guidelines proposed treatment for everyone over the age of 50 years.

The 2012 CTT meta-analysis reported significant reduction in major vascular events for patients with less than 10% 5-year risk, averaging 2.6% 5-year risk for major coronary events.

Two editorials that accompanied publication of the 2012 CTT meta-analysis in the Lancet commented on the benefit of statin therapy in low risk people found in the 2012 CTT meta-analysis. One cautioned about the increased risk of new onset diabetes, but stated “a good case could be made for treatment of individuals with an absolute risk of a cardiovascular event of less than 5% during 5 years with statins.”[5]

The other editorial, titled “Statins for all by the age of 50 years?” discussed the translation of the CTT findings into practice: “Because most people older than 50 years are likely to be at greater than 10% 10-year risk of cardiovascular disease, it would be more pragmatic to use age as the only indicator for statin prescription…” Both authors of this editorial were members of the Cochrane Heart Group.[6]

The 2013 Cochrane Review to which we refer concluded: “The individual patient data meta-analyses now provide strong evidence to support [statin] use in people at low risk of cardiovascular disease.” No lower limit of benefit was indicated.

The ACC/AHA cholesterol guidelines, issued three weeks after our Analysis was published in BMJ, recommend statin therapy for people with ≥ 7.5% 10-year risk of cardiovascular disease, with the option of statin therapy for people whose 10-year risk is between 5 and 7.4%. This further strengthens the argument presented in the 2012 editorial that the decision to prescribe statins could more cost-effectively be made simply on the basis of age. A 50-year-old American man with all values entered into the AHA risk calculator[7] at the 50th percentile has a 6% 10-year risk of cardiovascular disease. A 55-year-old American man with risk factors at the 50th percentile has a 10% 10-year risk of cardiovascular disease.

The updating of the evidence base resulted in an expected narrowing of confidence intervals and the addition of JUPITER trial added important evidence on diabetes risk.

Narrowing of the confidence intervals is, obviously, expected with the addition of clinical trial data. However, the effect of statin therapy on cardiovascular risk did not change, and therefore did not provide a basis for the Cochrane Review to radically change its recommendation from 2011 to 2013. Furthermore, we question the validity of the JUPITER trial data given that the trial was prematurely stopped at a time when there was no difference in cardiovascular mortality or serious adverse events between the rosuvastatin and control groups, and the unrealistically low mortality rate (8.8%) associated with myocardial infarction in the control group, compared to a 29% case fatality rate in the rosuvastatin group, raised questions about bias in event ascertainment.[8]

The authors consider that statins should reduce total mortality if they are to have a place in primary prevention in people in lower CVD risk strata and estimate a relative risk of 0.95, 95% confidence interval 0.86 to 1.04. However, the authors included both individuals with and without prior vascular disease in this estimate.

The “Main results” of the 2013 Cochrane Review reported: “Recent findings from the Cholesterol Treatment Trialists study using individual patient data meta-analysis indicate that these benefits are similar in people at lower (< 1% per year) risk of a major cardiovascular event.” The Table in our Analysis, titled “Comparative all cause mortality for low risk patients in statin studies included in Cholesterol Treatment Trialists’ meta-analysis” addresses this issue. As stated in our Analysis, statin therapy does not reduce over-all mortality significantly for the population of low-risk patients included in the clinical trials.

In considering the effect of statin therapy on the low-risk population, no clinical logic supports the removal of low risk patients diagnosed with cardiovascular disease from this calculation. Inclusion of these patients would be expected to increase rather than decrease the efficacy of statins. The discrepancy between the effects of treatment on the primary prevention population on the one hand and the low risk population as a whole on the other, exists because of a statistical quirk in the clinical trial data: low risk people (< 20% 10-year risk) with a history of vascular disease had no reduction in overall mortality (relative risk of overall mortality associated with statin therapy for < 5% risk = 1.04 and for ≥ 5% to < 10% = 1.00).[9] “Slicing and dicing” the data for overall mortality in the low risk group exploits the removal of these data to increase the estimated effect of statins.

Although the title of our Analysis addressing potential benefit of statin therapy in low risk people is correct, we agree that the précis beneath the title incorrectly points readers toward the effect of statin therapy in the primary prevention population rather than low risk population as a whole.

…the number of total deaths was small (1% of control group participants dying over 4 years) and non-CVD causes of death exceeded CVD deaths by more than 2:1… No strong evidence of benefit for total mortality was seen because other causes of death make up a greater proportion of total deaths, and it is unlikely that taking statins influences these non-CVD deaths.

We agree that, in the lower risk groups, by definition, cardiovascular causes of death will constitute a smaller proportion of total mortality. However, among low risk patients included in the CTT meta-analysis, statins failed to achieve a reduction in overall mortality and also failed to achieve a reduction in serious adverse events, and therefore no “net” or overall health benefit can be claimed.

We disagree with their statement that the “best indication of the net effect of a treatment on overall health is the total number of serious adverse events — which include deaths from all causes, hospital admissions, prolongations of admission, cancer, or permanent disability.”

We agree with this criticism. This sentence should read: “After overall mortality, the best indication of the net effect of a treatment on overall health…”

While criticizing the RCTs, they use low quality evidence from observational studies to support their statements about the hazards associated with statins despite likely high risk of bias.

We agree that observational studies provide a lower level of evidence than well-conducted RCT’s. Unfortunately none of the clinical trials included in the CTT meta-analysis compared drug therapy to lifestyle intervention in order to answer the question of how to most effectively and efficiently reduce the burden of cardiovascular disease on the low risk population. Thus, we relied upon the best available information.

A recently published metaepidemiological study included randomized controlled trials comparing the effectiveness of exercise and drug interventions in the reduction of mortality for secondary prevention of coronary heart disease, stroke, prediabetes, and congestive heart failure.[10] Results showed that exercise and statins were equally effective in patients with coronary heart disease, and exercise was 90% more effective than anticoagulants and antiplatelet drugs for stroke patients.

A publicly funded randomized controlled study comparing statin therapy, lifestyle intervention, and both for the prevention of cardiovascular disease in low risk patients would provide important comparative effectiveness information.

The authors also conflate muscle pain (myalgias), an important side effect of statins, with myopathy, a rare and more serious problem…

This is a semantic criticism, with which we disagree. From a practical point of view, statin-induced myopathy includes: myalgia (muscle symptoms without elevation in CK), myositis (elevated CK with or without muscle symptoms), and rhabdomyolysis (CK levels > 10 times the upper limit of normal).[11] Further, histopathologic findings of myopathy occur in patients with or without muscle symptoms and normal CK levels.[12] [13] [14]

The authors cite studies demonstrating adverse events associated with statin therapy but fail to cite systematic reviews which show no increased risk of psychological outcomes, fractures, acute renal failure, arthritis, or venous thromboembolism.

To narrow this concern, we did not comment on the relationship between statin therapy and the risk of fractures, arthritis, or venous embolism. In fact, an article we cited, based on 225,000 new users of statins in England and Wales, showed no significant association between statin therapy and these three adverse occurrences.[15] The same article did, however, show significant increased risk of acute renal failure associated with statin therapy. Because of under-ascertainment of adverse effects of statins in clinical trials, epidemiological data—like the data from general practices in England and Wales and the data addressing psychiatric adverse reaction from the New Zealand Center for Adverse Reactions Monitoring must be heeded as substantial warnings of potential adverse events.[16] The problem of under-ascertainment of adverse events in clinical trials is demonstrated by a 2006 meta-analysis of clinical trials involving statins that reported no increase in the risk of myalgia.[17] We agree that a more extensive review of the literature about each of these potential adverse reactions should include the full range of published articles.

Finally, Abramson et al. set up a false dichotomy, stating: “Rather than being compelled by guidelines to prescribe statin therapy for people at low risk of cardiovascular disease, doctors would provide a far greater service by explaining the magnitude of the benefits and uncertainty about the harms of statins together with discussion of the epidemiological evidence showing that behavioural risk factors—including tobacco use, lack of physical exercise, and unhealthy diet—are responsible for 80% of cardiovascular disease.” If they (and the BMJ editors) had awaited the publication of the 2013 ACC/AHA cholesterol guidelines, they would have been directed to the companion lifestyle guidelines, which aim to address these topics.

We agree that the 2013 ACC/AHA cholesterol guidelines include guidelines for positive lifestyle modification—as have the preceding guidelines. But once again, it is the expanded criteria for prescribing statins that is getting the attention, not the fundamental importance of lifestyle modification. The pro forma nature of the lifestyle recommendations is seen in an editorial written by a Vice Chair of the Expert Panel responsible for the 2013 ACC/AHA Guideline on the Treatment of Blood Cholesterol: “Despite decades of exhortation for improvement, the high prevalence of poor lifestyle behaviors leading to elevated cardiovascular disease risk factors persists, with myocardial infarction and stroke remaining the leading causes of death in the United States. Clearly, many more adults could benefit from evidence-directed use of statins for primary prevention.”[18] Once again, doctors are implored to “get real”--stop hoping that efforts to assist their patients and communities in the adoption of healthy lifestyle habits will succeed, and start prescribing more statins. This is a self-fulfilling prophecy. Note that the author of these comments disclosed receipt of funding from 11 drug companies, at least 4 of which are producing or developing new classes of cholesterol lowering biologics[19] that are projected to achieve annual sales of up to $10 billion per year.[20]

Our response to the Cochrane Reviewers’ comments would not be complete without addressing the evidence upon which they have relied. The Cochrane review adhered to strict standards of data collection and analysis, reviewing data for risk of bias, examining heterogeneity, sensitivity analysis, etc. However, there is no evidence they were granted access to the fundamental clinical evidence that prevents bias: patient level data from the clinical trials. The References section of the 2013 Cochrane Review shows that the source of data for every study included in the review was “published data only.”[21] Unlike the Cochrane Review, the Cholesterol Treatment Trialists Collaboration has been granted access to and relies upon patient-level data for every study incorporated into its meta-analyses.

In 2009, when the Cochrane Reviewers of neuraminidase inhibitors for prevention and treatment of influenza in healthy adults were unable to gain access to primary data from the manufacturer’s clinical trials of oseltamivir (Tamiflu),[22] they withdrew previous reviews.[23] [24] Access to patient-level data is similarly necessary for Cochrane reviewers of statin therapy to perform a thorough and independent evaluation. At the very least, analyses of prespecified outcome measures could be performed independently and the question of whether statins reduce the incidence of serious adverse events across all trials could be addressed.

If the Cochrane Reviewers have requested and been denied access to the patient-level data from the statin studies, we believe they should do what the reviewers of oseltamivir have done: publicly declare their inability to perform a responsible evaluation and retract conclusions that are based on published data only. The entire world is relying upon third party representations of data that the manufacturers should make readily available to Cochrane Reviewers and other academic researchers.

References
1 Abramson JD, Rosenberg HG, Jewell N, Wright JM. Should people at low risk of cardiovascular disease take a statin? BMJ 2013; 347:f6123. doi: 10.1136/bmj.f6123
2 Stone NJ, Robinson J, Lichtenstein AH, Bairey Merz CN, Lloyd-Jones DM, Blum CB, McBride P, Eckel RH, Schwartz JS, Goldberg AC, Shero ST, Gordon D, Smith Jr SC, Levy D, Watson K, Wilson PWF, 2013 ACC/AHA Guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults. J Am Coll Cardiol 2013; in press. doi:10.1016/j.jacc.2013.11.002.
3 Taylor F, Huffman MD, Macedo AF, et al. Statins for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev 2013; 1:CD004816. doi:10.1002/14651858.CD14004816.pub14651855
4 Cholesterol Treatment Trialists’ Collaborators. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380:581-590. doi:10.1016/S0140-6736(12)60367-5
5 Watts GF, Ooi EM, Balancing the cardiometabolic benefits and risks of statins, Lancet, 2012;380:541-3
6 Ebrahim S, Casas JP, Statins for all but the age of 50 years? Lancet, 2012; 380:545-7.
7 American Heart Association Heart Attack Risk Assessment, https://www.heart.org/gglRisk/main_en_US.html Accessed December 14, 2013
8 De Lorgeril M, Salen P, Abramson J, et al, Cholesterol Lowering, Cardiovascular Diseases, and the Rosuvastatin-JUPITER Controversy, Arch Intern Med. 2010;170(12):1032-1036
9 Webfigure 9: Supplement to: Cholesterol Treatment Trialists’ (CTT) Collaborators. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; published online May 17. DOI:10.1016/S0140-6736(12)60367-5.
10 Naci H, Ioannidis JPA, Comparative effectiveness of exercise and drug interventions on mortality outcomes:
metaepidemiological study, BMJ 2013;347:f5577 doi: 10.1136/bmj.f5577 (Published 1 October 2013)
11 Fernandez G, Spatz ES, Jablecki C, Phillips PS, Statin myopathy: A common dilemma not reflected in clinical trials, Cleveland Clinic Journal of Medicine, 2011;78:393-403
12 Draeger A, Monastyrskaya K, Mohaupt M, et al, Statin therapy induces ultrastructural damage in skeletal
muscle in patients without myalgia, Journal of Pathology, 2006; 210: 94–102
13 Association between statin-associated myopathy and skeletal muscle damage, Canadian Medical Association Journal, 2009;181(1-2):E11-E18
14 Phillips PS, Haas RH, Bannykh S, et al, Statin-Associated Myopathy with Normal Creatinine Kinase Levels, Annals of Internal Medicine, 2002;137:581-585
15 Hippisley-Cox J, Coupland C, Unintended effects of statins in men and women in England and Wales: population based cohort study using the QResearch database, BMJ 2010;340:c2197 doi:10.1136/bmj.c2197
16 Tatley M, Savage R. Psychiatric adverse reactions with statins, fibrates and ezetimibe. Implications for the use of lipid-lowering agents. Drug Safety 2007;30:195-201.
17 Kashani A, Phillips CO, Foody JM, Wang Y, Mangalmurti S, Ko DT, et al. Risks associated with statin therapy: a systematic overview of randomized clinical trials, Circulation 2006;114:2788-97.
18 Robinson JG, Accumulating Evidence for Statins in Primary Prevention, Journal of the American Medical Association, 2012;310:2405-6
19 Hatsko MAR, Are These Drugs the Future of High-Cholesterol Treatment? The Motley Fool,
http://www.fool.com/investing/general/2013/08/20/are-these-drugs-the-fut... Accessed December 17, 2013
20 Garde D,Amgen's PCSK9 drug hits PhIII mark in the race for a new cardio blockbuster, Fierce Biiotech, http://www.fiercebiotech.com/story/amgens-pcsk9-drug-hits-phiii-mark-rac... Accessed December 17, 2013
21 2013 Cochrane Review, pp 15-19
22 Jefferson T, Doshi T, Thompson M, Heneghan C, Ensuring safe and effective drugs: who can do what it takes? BMJ, 2011;342:148-51
23 Doshi P, Jefferson T, Del Mar C (2012) The Imperative to Share Clinical Study Reports: Recommendations from the Tamiflu Experience. PLoS Med 9(4): e1001201. doi:10.1371/journal.pmed.1001201
24 Jefferson T, Jones MA, Doshi P, Del Mar CB, Dooley L, Foxlee R. Neuraminidase inhibitors for preventing
and treating influenza in healthy adults. Cochrane Database of Systematic Reviews 2010, Issue 2. Art. No.: CD001265. DOI:10.1002/14651858.CD001265.pub3.

We thank Amrit Takhar for his thoughtful question regarding the final bullet point in the box titled “What low risk patients need to know.” The bullet states: “The side effects of statins…occur in about 20% of people treated with statins.” Dr. Takhar questions the validity of this statement because 35% of those who discontinued statin therapy due to adverse event were rechallenged with 92% of these patients continuing statin therapy 12 months later. We agree with this statement, but disagree with the opinion that we have “overstated the case against” statins.

First, as stated in the article by Zhang et al “The rate of reported statin-related events to statins was nearly 18%” in this retrospective cohort study. Second, the incidence of statin-related adverse events reported in the study is far more likely to be a floor rather than a ceiling. As noted in our article, spontaneous reporting of side effects is likely to underestimate the true incidence compared to rates determined prospectively by structured interview. Also, Zhang et al note, the incidence of side effects may have been under-reported because only the first reported statin-related event for each patient was included their analysis.

From a clinician’s perspective, the most important response to Dr. Takhar’s concern is that the incidence of statin-related side effects reported by Zhang et al was, in fact, “approximately one fifth.” For low risk patients who do not derive an overall benefit from statin therapy, the finding that many of the patients who experienced statin-related side effects could tolerate statin therapy on rechallenge does not negate the fact they experienced a drug-related side effect while taking a drug that provides them with no net health benefit.

John Abramson
Harriet Rosenberg
Nicholas Jewell
James M Wright

John Abramson and colleagues (Oct 26) provide a useful review and raise more uncertainty about whether the benefits of statin in those at low risk are sufficient to justify their use, The CTT study, which supports their use even in low risk patients, recorded muscle problems in 0.05% of the patients included (0.5/1000 treated with statins), raising concerns that the study did not document all those with symptoms as they quoted a large study of 13626 persons on statins compared with 32623 controls when the incidence of muscle problems were 100 times greater in a statin treated group i.e. 1 out of every 19 patients.

In clinical practice any musclulo-skeletal pains or other symptoms in statin users are a diagnostic and management problem, whether at high risk or low risk of cardio-vascular events. Is the symptom statin-induced or is there some other cause? The use of a statin free period has been suggested to evaluate any possible relationship. When a statin-free period is to be trialed it would be useful if packs of commonly presribed statin-like tablets were available with similar packs containing active but unlabelled statin available. These would enable a practioner to try a statin-free and a non-statin free period without patient or doctor knowing which is being taken i.e double blind. To add sophistication the patient could keep symptom scores and to add further sophistication statistical analysis of the results of the symptom scores using n=1 methods - which are available for time-series analysis (Knapp, MS, Using clinical evidence. Randomised controlled trials are not the only evidence. BMJ Jul 2001; 323: 165).

Providing packs for a double-blind trial of statin-cessation in individual patients seen in routine clinical practise, with a kit for the analysis of data collected, could be a service from the pharmaceutical companies who market statins to prescribers and patients.