A Novel Way To Think About Primary Prevention, Death, And Statins

A new paper offers a novel way to think about disease prevention.

The way we think– and talk– about disease prevention now is complicated and confusing. A key ingredient in the decision to take statins for the prevention of cardiovascular disease, for instance, is the estimation of 10-year cardiovascular risk. This single number, which represents the average of people with similar characteristics and risk factors, is incomprehensible or personally meaningless to nearly all patients as well as many of their healthcare providers.

Now the new paper, published in Open Heart and using the example of statin therapy, proposes a fresh and perhaps more meaningful way to consider this issue. The paper presents a framework for healthcare workers and patients to consider both the full range of the magnitude of benefit from statins and the likelihood of those gains.

UK researchers, led by Darrel Francis (National Heart and Lung Institute, UK), analyzed available data and calculated the likely range of increase in lifespan for individuals with similar cardiovascular risk profiles at baseline. Some of their findings are counterintuitive and unexpected. Although it is well known that risk increases with age, and that the absolute benefit of risk reduction treatments therefore increases with age, the authors emphasize that this does not translate into a greater increase in lifespan for people as they get older. “In fact, for any combination of cardiovascular risk factors, the potential lifespan gain from initiation of intervention decreases with increasing age of initiation. The gain for initiation at age 50 is approximately twofold to threefold larger than the gain for initiation at age 80.” In other words, younger people get the largest increase in lifespan, despite the fact that they are at lower risk when they start therapy.

Another key finding is that the great majority of people taking statins “gain no lifespan, while the minority that do gain, gain much more than the group average increase in lifespan.” The authors give the example of a 50-year-old, non-smoker, non-diabetic man with average cholesterol and blood pressure. On average starting a statin brings a mean life expectancy gain of only seven months. However, this can be viewed an entirely different way: 93% will have no gain in lifespan but 7% will have an enormous gain of 99 months.

This alternate perspective has important implications. As part of their study the authors conducted interviews with a representative sample of people in the UK. Their results show that people have different opinions about risk and that “it would be worth asking them their view individually.” As people are not computers, they “do not always choose the option offering the mathematically maximal average benefit.” Instead, the authors found, “our survey illustrates that people often prefer a small chance of a large benefit over the certainty of a small benefit, even when the mathematical average gain from the former is smaller.”

In an interview, Francis said that his group was “surprised to learn that many people value more highly a small chance of a big increase in lifespan than a certainty of a modest increase.”

A number of experts I contacted raised questions about the precise assumptions and models used in this paper, but I think they miss the point. The paper doesn’t offer a fully developed new framework for the consideration of primary prevention strategies. Instead it is intended as a starting point for a new and different model of thinking about risk and communicating with patients. Because our current methods are so abysmal, the paper is valuable for offering a provocative new way to think about these problems. It should be considered the start, not the end, of an interesting and important discussion.

I asked Saurabh Jha (University of Pennsylvania) for a comment on the paper. Here is his full response:

The gist of the paper is that not everyone with the same risk profile benefits equally from primary prevention and that the earlier the prevention is started the more life years there are to gain. The minority within a same-risk cohort that gain the most, gain a lot more than the average gain in life expectancy for the cohort. The average conceals two wings – those who gain a lot and those who gain little. Primary prevention mimics our economic situation. Just as wealth concentrates in a few, so, too, do the benefits of statins.

To give you an idea consider a 50 year old woman who has a national average risk of CV mortality and is on primary prevention. The average life expectancy gained for this cohort is 3 months. Nearly 97 % of women in this cohort will have no gains. But 3 % of women will gain an average of 92 months. Note, there are 28 times as many people who don’t gain than those who do gain. But those who gain, gain 24 times more than the average. Thus, following the average blindly leads to overtreatment of many. Ignoring the average leads to undertreatment of a few.

The researchers used a model. They populated the model with point estimates (which are, it must be emphasized, averages) from studies looking at risk reduction in cardiovascular mortality from primary prevention, and also life expectancy tables. They explored risk preferences by interviewing people. The model gave a spectrum of probabilities for life-expectancy gains for each individual on primary prevention using a Monte Carlo simulator (MCS). MCS is like throwing multiple dices at the same time – it is astrology with a dice.

As we age, the risk of dying from MI increases, the reduction in the risk of dying from MI (both absolute and relative) because of primary prevention also increases, but the gain in life expectancy because of primary prevention reduces. That is statins at age 80 are more likely to reduce your chances of death from MI than statins at age 50, but the gain in life years is less at age 80 than at age 50. The authors call this the “age paradox.” This is not really a paradox. It just means we don’t live forever, and the law of diminishing returns kicks in. The key point is that risk per se does not reflect the life expectancy gains from primary prevention as much as age. This complicates things because age is a marker of risk.

The study is a model which has limitations because a model must make assumptions. Nevertheless it makes a good point that the average does not apply to the individual. This is, in fact, the rationale for precision medicine. If we had perfect information we’d find that the benefits of primary prevention are even larger in fewer individuals (think Bush tax cuts). But in making this point, the researchers have used estimates, such as hazard ratios, which are also averages, and these averages were used to create guidelines which then present an opportunity for clinicians to undertake shared decision making with estimates from this paper. It’s like having your average and not eating it. There’s no getting away from averages in evidence-based medicine.

The study makes a reasonable case against using short-term horizons, such as 10 years, for adjudicating benefits. We should use the entire lifespan of an individual. The study, thus, also makes a case against using cause-specific mortality, such as deaths from cardiovascular disease, as a metric for success of therapy, as this disproportionately favors treating older, high risk people. If statins are to be started earlier, they should, by the same metrics, be stopped earlier.

The implication for shared decision making is trickier. Should you tell people that they have a small chance of a large effect – i.e. winner takes all? If so, should you also tell people that they have a large chance of gaining nothing? The answer is not straightforward, even though some may prefer small chances of large gains than a certainty of a small gain. The reason it is tricky is that preferences are influenced by framing. And how you frame the gains of statins depends on whether you want the person to be taking statins.

.

Curated comments: