Reading the evidence closely — statins for seniors

Recent news has reported that new research offers evidence for the benefits of taking statins for the elderly and for women — two groups of people in which statins have been especially controversial and not widely prescribed. Today, we’ll look at this new study on the use of statins in seniors.

To help us make the soundest health decisions for ourselves or a loved one, it’s first crucial to understand how clinical studies are reported. Oftentimes, the way results are reported don’t mean what we think they do.

And the mainstream media rarely gives the types of in-depth analyses of studies and the full information for us to be able to make informed decisions. Most of the news presents statins as if the science is settled and incontrovertible (there’s no need to repeat those stories here). So, many people may be surprised to learn of the medical debates that don’t regularly make the news.

Understanding risks — relative versus absolute

The largest body of evidence behind the use of statins has come from studies done on middle-aged men, rather than seniors or women. An editorial in today’s British Medical Journal gave a timely and helpful look at the difficulties for doctors and patients in sorting out the benefits and risks for statins, even for this most widely prescribed group of people. Dr. Des Spence from Glasgow wrote:

Whether it’s worth treating high cholesterol is a common enough question. No one who sees the charts and listens to the sales pitch [and media] would doubt it — but numbers are open to being spun. Let’s consider the trial known as WOSCOPS... The participants were men aged between 45 and 64 in the most socially deprived area in western Europe. More than three quarters (78%) were current or former smokers, and their average cholesterol concentration was 7 mmol/l [271 mg/dl]. If lowering of cholesterol concentration was going to work anywhere it was going to work here. The study ran for five years, and the researchers reported a 32% reduction in cardiovascular mortality in the group of men who took statins....

But, this impressive-sounding figure is a relative risk reduction, not the reduction seen in actual (absolute) heart disease deaths. So a woman might, for instance, lower her actual risk of dying from heart disease from 0.5% to 0.34% — by a mere 0.16% — not nearly as impressive as 32%! The real (absolute) numbers are important to know.

Dr. Spence went on to explain that for the men in the WOSCOPS trial:

Converting the 32% relative risk reduction into an absolute reduction gives a derisory 0.7% reduction in cardiovascular mortality and a number needed to treat of 143 over the study period.... So, putting it crudely, some 714 patients a year gain no benefit from treatment, even in the highest risk population in the world.... This is the “treatment paradox": that an individual patient, despite many years of investment in taking statins, gets virtually nil health benefit. Any relative benefit is seen only at the population level, even for composite cardiovascular endpoints. The treatment paradox is true of all treatable risk factors such as hypertension and osteoporosis.

Dr. Malcolm Kendrick, author of The Great Cholesterol Con, put these numbers into an even more meaningful perspective for an individual man. First, he cautioned that it’s easy to misinterpret that “numbers needed to treat” figure as suggesting a cure. “Statins do not treat/cure death, they only delay it,” he wrote. Assuming that you do gain one entire extra year of life for every 700 years of taking a statin, he explained, this means:

[I]f you treated someone for 30 years you can expect to provide them with 30/700 added years of life. This is 15.64 days... In short, if a fifty year old man asked you how much longer he could expect to live if he took a statin for thiry years you can inform him “just over two weeks — max.”

The rationale for treating elderly people with statins, however, is even “less clear because the association between plasma cholesterol and risk of coronary artery disease diminishes with increasing age,” said Dr. James Shepherd, M.D. and colleagues at the University of Glasgow, Royal Infirmary, Scotland. In fact, they said, “in the oldest old people, low plasma cholesterol is associated with increased mortality.” This has been recognized in the medical literature, they said, (examples here, here and here) for some two decades, while not widely known among the public.

This seeming paradox led these researchers to conduct the largest ever randomized controlled clinical trial of statin use in older seniors, at the greatest risk for dying and developing cardiovascular disease and strokes. The PROspective Study of Pravastatin in the Elderly at Risk (PROSPER) trial randomized 5,804 seniors, 70-82 years old, to a pravastatin (statin) or placebo group and followed them for 3.2 years. The results were published in a 2002 issue of Lancet. The study was popularly reported as supporting a benefit of statins in reducing coronary heart disease-related deaths for the elderly and the results reported in terms of a 24% reduction in relative risk. The actual rates of heart disease deaths, however, were 3.3% among the group on statins, compared to 4.2% among the placebo group — an actual (absolute) difference of only 0.9%.

Yet even this absolute risk was not the full story. This is only one of several wordsmithing things to pay close attention to when we read about the results of a study, so that we can understand what is being reported and what it means for us.

Meaningful endpoints

The endpoint that people most care about is whether the treatment actually prolongs life, which makes total mortality the most important consideration, not just heart disease mortality. The PROSPER trial found that while there were slightly fewer deaths from heart disease, there were more deaths from other causes, such as cancers and strokes. The all-cause mortality in this study differed by only 0.2% (10.3% in statin group and 10.5% in placebo control group). This is a clinically insignificant difference, and led three dozen professionals to write the National Institutes of Health to raise serious concerns about the lack of scientific evidence behind recommendations for statin use in elderly patients.

When you read statin studies, more commonly their benefits are described as reducing risks for surrogate endpoints, such as lab values, or for nonfatal cardiovascular events and unstable angina, and all-cause mortality isn’t mentioned at all.

Compounding the difficulty of determining if statins benefit the elderly has been the limited number of studies on them. Prior to the publication of the PROSPER trial in 2002, Dr. Scott Grundy, M.D. wrote an editorial for Archives of Internal Medicine saying: “To date, published clinical trials of statin therapy have not specifically targeted older persons... Without question, the lack of clinical trials that were designed specifically to test efficacy in older persons has been one factor standing in the way of a more forceful recommendation for intensive cholesterol-lowering therapy in this age group.” He argued for statin use in seniors 65 to 75 years of age, however, based on a subgroup analysis of seniors among the particpants other large trials which “strongly suggests that statin therapy significantly reduces risk in the age group.”

Dr. Uffe Ravnskov, M.D., Ph.D., a cholesterol researcher from Lund, Sweden, and colleagues, disagreed. As they explained, rather than relative risks, a more “honest way to inform the patient is to calculate his/her chance of surviving with and without treatment.” The personal odds for someone to benefit from any preventive drug or treatment is what people care about most: “What’s in it for me?”

They examined the actual chances of surviving, along with the relative risks and the absolute (actual) change in mortality, in the five trials Dr. Grundy used, as well as the large EXCEL (Expanded Clinical Evaluation of Lovastatin) double-blind, diet- and placebo-controlled trial of 8,245 seniors that he’d not included. They found that “the optimistic figures for relative risk reduction actually reflect unimpressive absolute risk reductions. These small benefits are also illustrated by the trivial differences between the survival rates with and without treatment.”

They compiled one of the clearest illustrations of the importance of looking at actual risks in order to gain a better understanding of what study findings mean:

They went on to examine other issues of statin use in elderly, such as mode of action and dosage. But they also highlighted yet another clinical consideration for anyone trying to weigh the risks and benefits of any medication. Just looking at mortality figures during a study period doesn’t consider potentially serious long-term side effects of statins, such as statin-induced myopathy, which is more frequent among the elderly, as Dr. Grundy cautioned. With advancing age, risk-benefits differ considerably from younger people and elderly are often more susceptible to adverse reactions. Drug metabolism is reduced in them and they are more likely to be on multiple drugs, to have multi-system conditions and to be underweight. According to Dr. Ravnskov and colleagues:

One may question whether statin treatment should be used at all because the small absolute risk reduction rewards may be outweighed by potential serious long-term side effects.

Whatever viewpoint you take away from the evidence, what is clear is that there appears considerable debate even among experts and each individual's decision may be different.

The latest evidence for statins in elderly

This past week brought news of a study purported to show statins, as a secondary prevention, reduce all-cause mortality in elderly heart disease patients. Researchers, led by Dr. Jonathan Afilalo, M.D., with McGill University in Montreal, Canada, published the results of their meta-analysis in the Journal of the American College of Cardiology.

Their working premise was: “Statins continue to be underutilized in elderly patients because evidence has not consistently shown that they reduce mortality.”

The authors searched through five electronic databases for trials published since 1966 that had examined statins use and included at least 50 elderly patients with documented cardiovascular disease. They identified 729 studies and winnowed them down, based on their inclusion criteria, to nine papers. The data in five studies they selected, however, had never been published and they did not obtain individual patient data from the authors, but extracted overall study data. Their inclusion criteria required each study to have data on all five outcomes: all-cause mortality, CHD mortality, nonfatal MI, need for revascularization and strokes. But the studies all had multiple different working hypotheses, used differing statins and doses, and were designed to examine different endpoints. Statin use among the control groups also varied from 2% to 24%. Women represented very small percentages of the study populations in each of their papers, but they believed that gender was not a significant effect modifier.

The Canadian researchers then used Bayesian forest plots with posterior relative risk estimates modeling for each study and pooled the relative risk estimates for five year follow-up periods. The five unpublished papers they used, however, had only followed the patients for 9 months to 3.9 years!

Already, we’re seeing some of the problems with meta-analyses, as has been previously reviewed. The studies lumped together in a meta-analysis can vary considerably in quality, measures, populations, methodologies and statistical analyses. Some studies may show a weak positive finding, others report none, and others may even report a negative finding. Meta-analyses can end up giving the same weight to all the studies (well-conducted and bad) by pooling them together, trying to create a statistically stronger estimation of an effect or prove something the studies weren’t designed to test. And therein lies the rub.

The Canadian study concluded:

In elderly patients with documented CHD, statins reduce all-cause mortality by 22% ... These estimates are rigorous and precise, owing in large part to our Bayesian hierarchical model and larger sample size of elderly patients, who had historically been under-represented in clinical trials. Achieving a high level of precision was critical, because summary odds ratios for all-cause mortality from 23 meta-analyses had been variable and heterogeneous.

This 22% reduction in relative risk, however did not reach statistical significance. In fact, their estimated pooled differences in absolute 5-year, all-cause mortality between the statin and placebo groups differed by only 3.1%.

In contrast to other studies, their model also showed increasing benefits from statins with age, even though none of the trials used in their meta-analysis included people over the age of 82, and the average age among the study populations was 69 years. Nevertheless, they concluded:

One of the most interesting findings of this study was that older patients attained a greater reduction than younger patients; the relative risk reduction for all-cause mortality was 50% in patients aged 80 to 97, 44% in patients aged 65 to 79, and 30% in patients aged 65 years.

They went on to conclude that their meta-analysis supported their premise and the need for doctors to change their statin prescribing habits for elderly patients:

Our meta-analysis adds to the current body of literature by showing that statins reduce all-cause mortality in elderly patients and that the magnitude of this effect is substantially larger than previously estimated....However, recently reported [statin] utilization rates of 40% to 60% in elderly patients with active CHD remain suboptimal. It is crucial to disseminate the evidence for statins in elderly patients with CHD to increase current utilization rates.

As the conclusions of the Canadian authors overstate the evidence they reported, one might wonder what, if any, conflicts of interest might be inadvertently influencing their analysis. But this published study had no Disclosure Statement, which is highly unusual for published studies.

An accompanying editorial in that same issue of the Journal of the American College of Cardiology discussed the problems that “continue to plague statin therapy in clinical practice: long-term adherence remains poor; and the treatment gap, especially among the elderly, remains large.” Doctors George A. Diamond and Sanjay Kaul of the University of California, Los Angeles, opined that the Centers for Medicare and Medicaid should discount drug prices to incentivize increased usage for those statins that show greater benefits, just as private insurers are already beginning to do. “A high discount will not erode the manufacturer’s profits,” they said. “In fact, profits are more likely to soar as a result of facilitated access to the less-costly drug and favorable shifts in market share.”

They said that statins are an example of the disconnect between what providers should do, according to the evidence, and what they’re paid to do:

This situation will not change unless and until we realign the financial and scientific incentives and begin rewarding caregivers, not for the prodigal provision of products and services, but for the enlightened provision of therapeutic benefit. Evidence-based reimbursement can be the bridge to this “far, far better thing."

[There was also no Disclosure Statement included in this editorial.]

Dilemmas for patients and doctors

When trying to decide what might be best for us or an elderly loved one, irrefutable evidence or at least all of the facts are needed so that we can weigh the risks and benefits. Yet other crucial information was missing from this recent meta-analysis and not considered in its conclusions. The Canadian researchers did not pool adverse events “because of failure to report these events stratified by age group in most studies and inconsistencies in classification of these events between studies.” So, they didn’t conduct a safety analysis to determine if the benefits of statin use outweighed the possible adverse events seniors might experience.

As recently reviewed here, evidence from younger populations for preventive health care interventions cannot be extrapolated to the elderly. Seniors are at special risk for adverse effects. One of the clearest examinations of the adverse side effects being reported among statin users and the implications for the care of elderly patients was written by Dr. Beatrice A. Golomb, M.D., Ph.D., with the University of California, San Diego. As she wrote in Geriatric Times:

[A]ll drugs have potential adverse reactions despite their potential benefits. Understanding these risks is vitally important, particularly in elderly patients in whom both risks and benefits differ relative to younger patients. Evidence suggests the balance of benefits to risks may be less favorable in the elderly: Cholesterol becomes a less potent predictor of cardiovascular problems, and adverse reactions from drugs, including statins, may become more prominent.

Her paper examined the most common statin adverse effects recorded in the University of California at San Diego Statin Study, including muscular problems, worsening cognitive function, gastrointestinal and neurological effects, psychiatric problems, immune effects (e.g., lupus-like syndrome), erectile dysfunction and gynecomastia (breast enlargement in men), rash and skin problems and sleep problems. She even cautioned against discarding the significance of higher incidences of cancers seen in the PROSPER trial of elderly patients based on the risks seen in studies of younger patients.

Making individual clinical decisions is never easy or as simple and one-size-fits-all as it can seem in the news. The low rates of statins being prescribed for elderly might not reflect doctors’ or patients’ lack of awareness of the evidence, as some have suggested, but a greater awareness. Perhaps doctors and patients have weighed the potential pluses and minuses and decided that, for them, statins were not best.

Will that choice continue to be an option, or will statins become mandated performance measures for doctors, as already seen in some places, and obligatory for patients in order to receive medical care without being labeled as “noncompliant?” The importance of the best evidence behind clinical care guidelines impacts healthcare in ways many people might never imagine. One reader shared the difficulties she was having in finding a doctor to care for unrelated health problems because of, she believes, her decision not to take statins. As she told JFS: “No physician is obliged by law to take me if I won’t comply with 'standard of care.' I must take statins or I have no doctor.”

© 2008 Sandy Szwarc