Death by FIRE?


I’m dying early, it seems. I didn’t realize it, but a recent Oregon State University study claims to show that people that retire early are “11% more likely” to die earlier than those that work past age 65.  Perhaps reaching FIRE (financial independence and retiring early) is bad for my health?

The 18 year long study (1992-2010) had ~2,500 participants and showed that 12% of healthy people starting the study died during the study time frame. Among these, early retirees were “11% more likely” to die before people that worked past age 65.  

Seems scientific, doesn’t it? The results are reported as a paradoxical cautionary tale against leaving the workplace.  At least they controlled for whether or not people were healthy (after all, health issues force many people into early retirement). Not all studies on this topic do so.  

Yet, a couple things bother me about this study: 

First, I don’t like it when researchers report a percentage of a percentage. It makes small differences look bigger than they are. If 12% of healthy people died during the study – how big a difference does an “11% higher risk of mortality” really look like? Well it looks like this: 12.4% of early retirees died, versus 11.2% of people who worked past 65. That’s a 1.2 percentage point gap.  It certainly seems pretty close to me. Stated another way, you would say 2 out of 20 early retirees died, as compared to 2 out of 20 people that worked until past age 65. No difference at all when you put it in these common terms, because the very slight difference represented in the study gets rounded off. 

Second, I’ve found no information cited on what the margin of error for a study like this might be. They are using ~2,500 participants to project to hundreds of millions of Americans. At what level of confidence are they projecting statistical significance? Political polls typically have 3-4 percentage points of possible statistical error. It’s hard to imagine the difference reported here isn’t also within the margin of error.  

Lastly, there are significant confounding variables that are not controlled for in the study results. Single people, for instance, tend to retire earlier while married people live longer. Self-employed people tend to work later, but live happier and longer. Men and women tend to retire at different ages (as do people with different ethnicity). People with higher levels of education tend to retire later than others. White collar workers retire later than blue collar workers and have jobs that are less physically demanding. Additionally, people that work longer would likely earn more money by working longer and wealth is highly correlated with living longer.  

These variables taken separately or together would seem to have significant bearing on a statistical difference that already seems relatively small. Small enough that I will certainly take my chances and instead focus on the things that are more obviously shown to make a much bigger impact on longevity: eating well, being active, and having strong relationships with my wife, family, and friends.  

Image Credit: Pixabay

26 thoughts on “Death by FIRE?

  1. This study is definitely missing some scientific rigor, the percentage difference are indeed very small.
    I think the degree of increased exercise, healthy food consumption and reduced stress can easily compensate the chances of early retirees to die earlier by 0.4%!

    Like

  2. Seems like a very small sample size. I believe the biggest factors no matter what age you retire is the things you mentioned, eating well, being active, and having strong relationships. For many their jobs define them, once the leave it they lose their purpose. It starts going down hill from there.

    Liked by 1 person

  3. Agreed. A study which may well have been conducted fine but reported out the data the wrong way completely. Often the case, unfortunately in science.

    Reminds me lot of why the FDA won’t approve new cardiovascular drugs easily without large and costly “outcome studies”. What this boils down to is showing superiority with the new drug versus placebo on a background standard of care. This require tens of thousands of patients to be followed closely for three or more years. Yeah, very very costly. Designing such studies and slicing the data has to done extremely carefully with the stakes being so high for the patients and the sponsoring company.

    Liked by 1 person

  4. Very small sample size for sure… and I wonder how many of these early retirees that passed away were forced to take an early retirement and how many of them tried to find a job afterwards but couldn’t get their feet in the job force.

    Liked by 3 people

  5. The study is straight out the junk science mill. As you stated, the probabilities are too close to really be called statistically different. As a rule of thumb, unless the additional risk factor at least doubles the risk relative to the unconditional probability, there is nothing there. 11% increase is insignificant.
    Another issue the researchers missed is the fact that for some professions the retirement date is early, BECAUSE the work is so taxing and back-breaking (coal miner, etc.) so the causality goes the other way around: they get early retirement deals because they likely have lower life expectancy.

    Liked by 1 person

  6. Agree completely. You raised the important question mark and variables that need to be covered . We need to see and know the assumptions that go into the study to truly understand if the study impacts us and if we fit into the category. I have no clue if I fit into this study or not. I tend to avoid these studies for the reason you mentioned.

    Plus, here is the question I want to know. For the people that died early, were they going to regardless of when they retired? Then, did retiring early allow them to experience stuff that they wouldn’t have gotten to experience because of the extra time?

    Bert

    Liked by 1 person

    • I was thinking of that same thought! Additionally, I wondered how much earlier they died then the folks that worked until they were at least 65? and, was that less than the amount of time they enjoyed not having to work? Lots of questions…

      Like

  7. Despite the (unproven) claims above of being a badly performed study (or statistically invalid), this study is most probably correct, as was one that was executed on former RDS (shell) employee’s which retired at age 55. The found that the primary reason was the lack of social interaction and drive in the morning to get going. Apparently they kind of died early from boredom and/or loneliness (i.e. mental issues). Mind you, this risk was still relatively small (as is also shown in this study with the 11%).

    Furthermore, this probably does not apply to those wanting to get out early (as in those striving for FIRE), since these folks (including ourselves) are planning on things to do once they are financially independent. Different attitude towards life after retirement.

    Like

    • Good thoughts. For the record, I did not claim it was a “badly performed study”. I am merely pointing out that the difference is very small, the level of statistical significance was not reported with the study findings, and that there are probably other confounding variables. I agree with your thoughts that the results may not apply to those seeking FIRE, as they are likely a distinct attitudinal segment with some characteristics that are positively correlated with longevity.

      Like

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s