A new 10-year mortality estimator — and its potential utility

As we age, and have to consider the risks and benefits of all sorts of health-related issues, it is often both helpful and important to have a clear and reasonable idea of how long we might expect to live. This information can help us make better decisions about everything from how much money we need to have when we retire to whether we really need to go on getting PSA tests when our PSA level is still 1.9 at age 69 (if we are overweight, have diabetes, and have already had one serious cardiovascular event).

A research letter by Cruz et al. in this week’s Journal of the American Medical Association (JAMA) has offered physicians a new “mortality index” — a way to assess the 10-year mortality risk for their patients. As a research letter, this article has no abstract, but there is  more information about this paper in a media release issued by JAMA and in a commentary on the HealthDay web site.

Basically, Cruz et al. set out to determine whether an earlier 4-year mortality index that their research team had developed — based on data from some 20,000 people who participated in the Health and Retirement Study (a nationally representative cohort of community-dwelling, U.S. adults older than 50 years) — could be used to project 10-year mortality with reasonable accuracy.

The mortality index is based on a scale that assigns patients points associated with 12 different health-related factors: age; sex (that’s M or F, not whether you are having any); current tobacco use, body mass index (specifically whether you have a BMI < 25); whether you are diabetic; whether you have (or have had) any cancer other than non-malignant skin cancer, chronic lung disease, or  heart failure;  whether you have difficulty bathing yourself, managing your finances (because of a mental deficit of some type), walking several blocks, or pushing or pulling large objects (e.g., an armchair).

Here is the actual scoring assignment chart:


By adding up an individual’s total point score, based on the above factors, the researchers were able to assign a total score (from a minimum of zero [for a woman who was slightly overweight but no other health-related conditions] to a maximum of 26 [for man ≥ 85 years of age with all of the problematic health issues]) and then a risk level (from 0 to 100 percent) for that individual.

The two slightly puzzling issues in this point assignment system are:

  • Why a person with a normal or lower than normal BMI would be at greater risk than a man who was slightly or significantly overweight. (This may be because older people with low BMIs may be relatively frail by comparison with those who have higher BMIs.)
  • Why there is no reference to risk related to significant substance abuse. (One would assume that, for example, heavy drinkers of alcohol were actually at greater risk of death, for all sots of possible reasons.)

In their development cohort, Cruz et al. showed that 10-year mortality rates ranged from a risk level of 2.5 percent (among 12/486 participants with 0 points) to 96 percent (among 298/310 participants with 14 or more points). Note that no one is at zero risk of dying within 10 years. We can all just get “hit by a bus.”

When they then tested the mortality index in a validation cohort, they found that 10-year mortality rates ranged from 2.3 percent (among 8/354 patients) to 93 percent ( among 239/257 patients).

The authors conclude that they had been able to validate “a mortality index that accurately stratified older adults into groups at varying risk for 10-year mortality.”

They go on to suggest that, “Patients identified by this index as having a high risk of 10-year mortality may be more likely to be harmed by preventive interventions with long lag times to benefit, whereas patients identified as having a low risk of 10-year mortality may be good candidates for such interventions.”

Clearly, a mortality index like this has some relevance not only to decisions about whether one should continue to have PSA tests at a point in time, but also to whether one should continue on active surveillance (as opposed to just watchful waiting) if one clearly has low-risk disease that is non-progressive if one has a high 10-year mortality risk.

Now we do need to be very clear that a mortality index like this is a statistical probability tool. It is not some sort of absolute indicator of risk for 10-year mortality for any one individual. On the other hand, it does have the potential to give an individual guidance. We also need to be clear that this tool is only valid for people of > 50 years of age.

The “New” Prostate Cancer InfoLink intends to find out whether this mortality index will become available as a on-line tool for patients, or whether access to this tool will be restricted to physicians. The latter would be regrettable — in our opinion.

One Response

  1. This paper and mortality model are a potentially useful development. Thanks for posting about them. I have one main concern — cancer’s 2 points.

    I’m thinking those 2 points assigned for any current or prior cancer cover a wide range of situations that would affect mortality predictions in much different ways. The authors did eliminate non-malignant skin cancer, but low-risk prostate cancer likely offers the same level of very low risk. In contrast, most pancreatic cancer still cuts life short quickly, and lung cancer is also a major life shortener. As prostate cancer is one of the most common cancers, but with no life curtailment for the vast majority of us and typically long survival for the rest, it would be helpful if someone with an actuarial bent would take this model the next mile and adjust the points for prostate cancer by risk level. The same thoughts would likely apply to breast cancer and colorectal cancer, among the major cancers. A full 2-point credit would apparently improve odds at 10 years by perhaps 10% (my eye view of the hazy graph curves) by reducing from 8 points to 6 points, for instance.

    I’m curious whether the authors considered these points at all or just took a national figure for overall cancer survival to plug into their results. I’m thinking the authors all come from the Geriatrics Division, San Francisco Veterans Administration Medical Center, UCSF, as, though unstated in the published letter, all the authors of the original 4-year mortality model came from that division. That medical community and culture specializing in general geriatrics would likely be important in constructing a model as such medical professionals would likely not be familiar with the uneven topography of cancer survival and risk levels within the common cancers. Here is the citation for the original paper on 4-year mortality:

    Development and validation of a prognostic index for 4-year mortality in older adults. Lee SJ, Lindquist K, Segal MR, Covinsky KE.
    JAMA. 2006 Feb 15;295(7):801-8. Erratum in: JAMA. 2006 Apr 26;295(16):1900. PMID: 16478903

    It is worth noting that cancer was also assigned 2 points in that 2006 paper; in other words, no change. While perhaps too subtle a distinction for a general mortality model, it is clear that there have been distinct improvements in cancer survival in just the years from 2006 to 2013, and improvements are likely to continue (even if at a slower rate if the NIH/NCI budgets are again reduced).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: