Request a Demo

Multidisciplinary Team

It’s Time to Reset the Scale on Severe Obesity

Connect on LinkedIn
Hero Image
Connect on LinkedIn

Most life insurance carriers won’t even consider applications with severe obesity. Turns out, though, that nearly half of them (we’re talking millions and millions of Americans with BMI >40) are good and even very good risks—and that this practice is costing you big business.

We get it: as a uniform cohort, relative mortality for those with severe obesity is 190%. Your guidelines are governed by the data you have access to, so this figure has certainly justified jet-declining severe obesity as a protective measure. But our ever-curious actuaries and clinicians have long believed that it’s possible to wrest viable underwriting opportunities from the clenched fist of BMI/build jet declines. We just needed the data to prove it.

And now we have it. Christian Shepley, Emily Simons, Kimberly Sapre, and Sue Bartholf dug deep into vast volumes of it and applied our sophisticated insurtech tools to uncover insights that will both surprise and delight you.

They found that after pulling out lives with a few key comorbidities and accounting for factors like a history of significant weight fluctuations, the remaining cohort has a low average risk score, indicating a low aggregate relative mortality despite their severe obesity.

Our researchers presented their findings in a blockbuster webinar, but the backstory is also fascinating. We ask them here how they went about mining the eminently insurable applicants from the forbidden risk terrain of severe obesity, enabling you to confidently write more business and cover millions more people.

The bottom line? Data-driven tools make it possible to clearly see individual risk in its unfathomable uniqueness and nuance, rather than through the foggy lens of generalized condition knowledge.

Meet the team

Is it true that this research was prompted by Irix Risk Score customers who were baffled when the model generated surprisingly low scores for some very high-BMI applicants?

Christian: I had wanted to investigate BMI and build data for a while because they’re such important factors in underwriting, but it became a priority when Risk Score clients questioned applicants with obesity and low Risk Scores indicating that they were good risks.

Carriers couldn’t reconcile those low scores with their preconceived notions, so, we set up a study to determine whether we needed to improve our model—or prove that Risk Score was accurately identifying healthy individuals in BMI ranges thought to be high-risk.

Ultimately, we showed that the model was correct, but even we were surprised how many people with severe obesity were otherwise healthy.

One thing people may notice is that in our webinar we talk about “people with obesity” as opposed to people who are obese. What’s behind that change?

Kimberly: In medicine we have moved to using what is called “person-first language.” So, you will hear “people with obesity” versus “obese people.” Another example would be to say, “people with diabetes” instead of “diabetics.” It is a way to put the person first and not their condition.

What’s wrong with jet-declining for severe obesity?

Kimberly: First, we usually diagnose obesity on the basis of BMI, a simple equation based only on height and weight. As a statistical tool, BMI is unable to distinguish body fat from muscle mass. And second, even in cases where BMI does provide a roughly accurate impression of a person’s percentage of body fat, determining whether the person is healthy or not depends on other factors, such as whether they’re physically active and where the fat is stored in their bodies.

What’s the role of the data analyst on a project like this? Are you going into the data to look for specific things—following up on your own or others’ suspicions—or are you looking for interesting statistical associations and bringing them to the team’s attention?

Emily: Being the analyst in a project like this is a mix between hunting down the answers to existing questions, following up on suspicions, and looking for statistical associations that we did not previously think about. Our suspicion was that BMI was not a strong predictor of mortality due to its lack of robustness as a measure of obesity. This informed our initial question of, Is there any relationship between BMI and mortality?

We found that yes, BMI does have a relationship with mortality, but it wasn’t quite what we expected. That led us down the path of searching for why that relationship was the way it was and how we could use our data to help stratify within a given build classification. We found that our data in combination with predictive modeling techniques can help underwriters appropriately use BMI, despite its shortcomings as a metric on its own.

What metric would be better than BMI?

Kimberly: There are a number of technological solutions that provide more accurate measure of body composition, such as smart scales and DEXA scans that can measure overall body fat percentage and where fat is stored, but those solutions are expensive and time-consuming. A good compromise would be to use BMI in conjunction with waist circumference, or to incorporate other calculations such as the waist-to-height ratio or waist-to-hip ratio. Such a solution would serve as an inexpensive, low-tech metric that would be an improvement over BMI alone as a measure of obesity and health.

One of your key findings was that the elevated mortality associated with severe obesity is largely driven by comorbidities and other factors like patterns of weight fluctuation. What would you say to a skeptic who’d argue that, given time, most people with severe obesity will develop comorbidities?

Christian: We looked as far back into our mortality data as we could, identifying a statistically significant population we could follow for 13 durations. What we found was that there was a degree of protective wear-off or regression to the mean, but it was similar to what we’d see in any other type of medical underwriting.

Predictive models like Risk Score allow us to stratify risk and identify good risks within conditions that carriers used to decline almost automatically, like severe obesity. What are the implications for an industry that has pooled risks for centuries?

Sue: We are not arguing against risk-pooling; after all, it is a fundamental principle of insurance and underwriting. We are talking about more accurately assessing the individual’s risk so that the risk-pooling is more accurate, which ultimately improves the matching of pricing and risk for the pool. That’s both good for business and, from the applicant’s perspective, fairer; being evaluated on your own merits is preferable to being lumped into a group and judged en masse.

If the use of Risk Score can take a segment of the population that carriers won’t even consider, like those with BMIs over 40, and show you that 48% of them are actually highly insurable, doesn’t that make you wonder what other opportunities you’re missing?

To learn more about our severe obesity mortality findings, view the full on-demand webinar.

"We are not arguing against risk-pooling; after all, it is a fundamental principle of insurance and underwriting. We are talking about more accurately assessing the individual's risk so that the risk-pooling is more accurate, which ultimately improves the matching of pricing and risk for the pool."

Interested in becoming a part of the IntelliScript family? We look forward to hearing from you.

View Open Positions