69色情片's analysis of the UK Biobank study unveils compelling correlations between lifestyle factors and mortality risk, offering insurers a fresh perspective on risk assessment. By integrating both traditional and novel biometric indicators, this research lays the groundwork for a shift in underwriting practices, potentially ushering in an era of more precise and holistic risk evaluation in the insurance sector.
Supported by a large body of published research, insurers have long understood the significance of features such as body mass index (BMI), blood pressure, and serum cholesterol as predictive risk factors for poor health outcomes.1-5
However, while these metrics are well-established indicators of health risks in the general population, insurers have often lacked the ability to accurately quantify the associations between these variables and the risk of mortality and morbidity events in insured lives. Bridging this gap is critical and can offer multiple benefits to both applicants and insurers.
For insurers, enhancing risk assessment can improve underwriting strategies, ultimately leading to optimized pricing. For applicants, a greater understanding of how factors such as exercise, influence their health could encourage behavior changes and strengthen engagement with insurance-related wellness programs, positioning insurers as partners in their pursuit of a healthy lifestyle.
While traditional risk factors such as BMI have been used universally for decades in underwriting, non-traditional risk factors such as walking pace and objective measures of physical activity remain relatively new and have yet to be fully integrated into underwriting practices. For example, despite 69色情片 identifying step counts as a meaningful predictor of mortality several years ago,6 most underwriting manuals have refrained from incorporating this metric due to a lack of relevant data in insurable lives.
In this paper, we share some valuable insights from our UK Biobank study with the University of Leicester into how traditional (e.g., BMI) and non-traditional (e.g., wearable-measured step counts) risk factors impact mortality.7,8 Since 2022, 69色情片 has been sponsoring and collaborating with world-renowned academics from the University of Leicester to develop valuable research insights from the UK Biobank database to help strengthen the understanding of biometric data for disease and death prognostication.7-9 Interestingly, due to a 鈥渉ealthy volunteer鈥 selection bias,10 the UK Biobank is naturally more representative of standard insured lives in comparison to the general population, making these unique insights particularly relevant to the insurance industry. We took it one step further by ensuring the risk models in our analysis were also:
Here, we segmented the UK Biobank database into the different cohorts before applying advanced statistical techniques to:
Finally, we examine the broader implications of this research and explore its potential value for insurers.
Figure 1 below illustrates the strong 鈥淯-shaped鈥 relationship between BMI and mortality risk in UK Biobank male participants under 60, after controlling for traditional underwriting risk factors. An individual with a BMI of 17.5 kg/m2 was found to have three times the mortality risk compared to a peer with a BMI of 27 kg/m2 (median BMI value in this age and gender group). Similarly, at the other end of the spectrum, the mortality risk in an individual with a BMI of 50 kg/m2 also had an approximately three times higher mortality risk compared to the median. This significant association was largely consistent across all data subsets (i.e., all age and sex groups as well as proxy for standard, rated, and chronic disease lives; data not shown).
The most striking result is the high mortality impact at low BMI values, particularly those in the range of 17.5 to 20 kg/m2. Not surprisingly, such findings have been underreported because previous medical studies have been focused on different populations or have employed different modeling strategies.
For instance, in relation to the latter, many have not fully controlled for underwriting-relevant risk factors or failed to appropriately address non-linearity in the data (e.g., categorized the data or not allowed for sufficient flexibility via optimizing the number and positions of knots that define the spline terms).
For example, Bhaskaran et al. conducted a population-level analysis of 3.6 million UK adults to investigate the impact of BMI on overall and cause-specific mortality.1 They observed a 鈥淛-Shaped鈥 relationship with a mortality ratio of around 2 at low BMI values. Similar findings were also presented in two large meta-analyses.2,3 More recently, Rigatti and Stout quantified the mortality effect of low BMI in life insurance applicants.11 However, they also reported smaller hazard ratios than those observed in the UK Biobank data.
Further research is needed to better understand the significance of these findings and their implications for underwriting individuals with low BMI. In particular, Sun et al. showed that an increased risk of mortality for being underweight was only evident in ever smokers in the UK Biobank as well as the HUNT study.12
Figure 2 shows the compelling association between self-reported walking pace and mortality risk in all UK Biobank male participants under 60, after controlling for traditional underwriting risk factors. Compared to individuals who typically walk at a slow pace, those with a steady average walking pace had a 40% lower mortality risk, while those who usually walk briskly had a 50% lower risk. This robust relationship remained consistent across all data subsets (i.e., all age and sex groups as well as proxy for standard, rated, and chronic disease lives; data not shown).
Although this association has previously been reported in the literature,13,14 it is remarkable to see it sustained in our various subsets of insurable lives. Moreover, incorporating self-reported walking pace into the base model for UK Biobank male participants under 60 increased the model鈥檚 predictive power for mortality risk by ~2%, demonstrating the additional value this metric provides beyond traditional underwriting risk factors. Again, this finding was largely consistent across data subsets (i.e., all age and sex groups, as well as proxy for standard, rated, and chronic disease lives; data not shown).
Figure 3 below shows the predictive power of replacing each traditional underwriting risk factor in the base model (i.e., BMI, smoker status, systolic blood pressure, and total cholesterol) with self-reported walking pace in all UK Biobank male participants under 60.
For example, the model鈥檚 ability to predict mortality risk increased slightly when systolic blood pressure was replaced with self-reported walking pace. A similar effect was observed for total cholesterol, suggesting that self-reported walking pace is a marginally stronger predictor of mortality than systolic blood pressure and total cholesterol in the UK Biobank cohort. However, model accuracy declined when BMI and smoker status were independently replaced with self-reported walking pace 鈥 emphasizing the importance of these traditional underwriting risk factors in the UK Biobank cohort. These findings were generally consistent across different data subsets (i.e., all age and gender groups, as well as proxy for standard, rated, and chronic disease lives; data not shown).
It is important to note that walking pace data in the UK Biobank were self-reported 鈥 not objectively measured. As a result, these results should be interpreted with caution.
Figure 4 illustrates the predominantly negative linear relationship between wearable-measured step counts and mortality risk in UK Biobank male participants under 60, after controlling for traditional underwriting risk factors.
For example, individuals averaging ~5,000 steps per day had an approximately 1.5 times higher mortality risk compared to those with ~11,000 steps per day (median step count value in this age group). Conversely, individuals with ~15,000 steps per day had an approximately 30% lower mortality risk compared to the median.
While the overall pattern was similar in adults aged 60 and older, the relationship appeared to be stronger in younger participants (data not shown). This association was generally consistent across all data subsets (i.e., proxy for standard, rated, and chronic disease lives; data not shown).
Compared to published results from a recent meta-analysis,15 our models, fitted to the various insurable populations, largely suggest that step counts have a slightly weaker impact on mortality. However, this is not surprising and is likely due to our inclusion of additional underwriting-relevant risk factors, which slightly attenuates the observed relationships.
Nonetheless, this remains an insightful and original finding, as this data comes from the largest objectively measured physical activity study in the world, with over 100,000 participants wearing wrist-worn accelerometers over a period of one week for 24 hours per day.
Figure 5 below shows the predictive power of replacing each traditional underwriting risk factor in the base model (i.e., BMI, smoker status, systolic blood pressure, and total cholesterol) with step counts in UK Biobank male participants under 60.
For example, the model鈥檚 ability to predict mortality risk increased by over 1% once total cholesterol was replaced by step counts. Similar effects were observed for BMI and systolic blood pressure, suggesting that step count is a stronger predictor of mortality than BMI, systolic blood pressure, and total cholesterol in the UK Biobank cohort. Model accuracy decreased when smoker status was replaced with step counts 鈥 suggesting that smoker status is a more important predictor of mortality in the UK Biobank cohort. Findings were indistinguishable across the different subsets of the data (i.e., older lives as well as proxy for standard, rated, and chronic disease lives; data not shown).
The findings from our UK Biobank study, in partnership with the University of Leicester, have clearly demonstrated that:
BMI plays a fundamental part in underwriting life and health insurance, while metrics such as self-reported walking pace and step counts have yet to be fully integrated into underwriting practices. Our findings regarding low BMI values are especially thought-provoking. Still, even though some insurers may already be aware of this, further research is needed before recommending broader underwriting changes.
On the other hand, as the industry increasingly seeks new and alternative data sources for underwriting life and health insurance,16 these findings provide compelling evidence on how self-reported walking pace and step counts can effectively segment mortality risk and complement traditional underwriting risk factors. These insights not only advance public health understanding of biometrics, healthy living, and mortality risks but also offer (re)insurers opportunities to refine underwriting philosophies and enhance wellness strategies.
In this paper, we have presented robust evidence on the prognostic value of BMI, self-reported walking pace, and wearable-measured step counts; findings that are highly applicable to insured lives. As the industry鈥檚 interest in biometric data continues to grow, these insights have the potential to enhance underwriting and wellness strategies, while also helping applicants make informed decisions to improve their longevity.
Further insights from our ongoing research and collaboration with the University of Leicester are expected to be released later this year.
with the lead researchers of the joint 69色情片 and University of Leicester study of UK Biobank data to learn more about quantifying biometrics risk factors.
