Background  
Studies have shown that increased BMI is associated with elevated mortality risks. This may be driven by case fatality of obesity complications that occur more frequently with higher BMI (disease mortality), or by residual mortality risk associated with higher BMI (BMI mortality). Previous versions of the recently validated Core Obesity Model (COM)(1,2) adjust for BMI impact on the complications of obesity, including prediabetes and type 2 diabetes, but assume no further impact of increasing BMI on mortality. 

  


Aim  
To assess the impact of different ways of modelling mortality on all-cause mortality in the COM (version 15).

Method  
All-cause mortality by age and sex, derived from general population statistics, is incorporated in the COM. To avoid double-counting, mortality associated with specific conditions such as cardiovascular disease, cancer and diabetes, which are accounted for in the modelling, is subsequently subtracted. Finally, obesity complications and mortality are estimated according to physiological parameters, including BMI. A targeted literature review using PubMed was conducted to identify recent studies (published 2010–2020) investigating the association between mortality and BMI in populations with obesity, in order to further calibrate mortality in the model.

Results  
The review identified six studies for potential inclusion, of which two large, population-based UK studies were used. Both studies used BMI 25 kg/m(2) as the reference category. Bhaskaran et al.(3) reported a hazard ratio (HR) of 1.21 for each five BMI units increase, and Wade et al.(4) reported a HR of 1.02 for each one BMI unit increase, above this reference category. Based on these two studies, four additional options to predict all-cause mortality (10-year risks) were included in the model, by calibrating with BMI (Table). When predicting mortality based on disease mortality only, the risk is fairly constant across BMI categories. However, accounting for combined disease and BMI mortality or BMI mortality alone yields greater differences in mortality risks across BMI categories.  
  


[ Table Removed ]

[ Table Removed ]

(a)Disease-specific plus non-disease-specific mortality (by age, sex and disease history, derived from general population statistics).   
Baseline characteristics: age, 48.5 years; 64.6% women.

Discussion  
The updated version of the COM allows the user to apply different ways of calculating mortality. Applying the impact of BMI only on disease-specific case fatality rates results in small differences between BMI categories, whereas applying the effects of BMI on general mortality results in a larger spread of mortality risks. Combining both approaches results in a wider spread of risks, without changing the mortality risks in the normal BMI group, aligned with the evidence base informing the calibration.

References  
1. Lopes S et al . Obes Sci Pract 2021;7:269–80.   
2. Lopes S et al . Pharmacoeconomics 2020;38:1123–33.   
3. Bhaskaran K et al. Lancet Diabetes Endocrinol 2018;6:944–53.   
4. Wade K et al. Obesity (Silver Spring) 2018;26:1796–806.



**Background** 
Studies have shown that increased BMI is associated with elevated mortality risks. This may be driven by case fatality of obesity complications that occur more frequently with higher BMI (disease mortality), or by residual mortality risk associated with higher BMI (BMI mortality). Previous versions of the recently validated Core Obesity Model (COM)(1,2) adjust for BMI impact on the complications of obesity, including prediabetes and type 2 diabetes, but assume no further impact of increasing BMI on mortality. 
**Aim** 
To assess the impact of different ways of modelling mortality on all-cause mortality in the COM (version 15).
**Method** 
All-cause mortality by age and sex, derived from general population statistics, is incorporated in the COM. To avoid double-counting, mortality associated with specific conditions such as cardiovascular disease, cancer and diabetes, which are accounted for in the modelling, is subsequently subtracted. Finally, obesity complications and mortality are estimated according to physiological parameters, including BMI. A targeted literature review using PubMed was conducted to identify recent studies (published 2010–2020) investigating the association between mortality and BMI in populations with obesity, in order to further calibrate mortality in the model.
**Results** 
The review identified six studies for potential inclusion, of which two large, population-based UK studies were used. Both studies used BMI 25 kg/m(2) as the reference category. Bhaskaran et al.(3) reported a hazard ratio (HR) of 1.21 for each five BMI units increase, and Wade et al.(4) reported a HR of 1.02 for each one BMI unit increase, above this reference category. Based on these two studies, four additional options to predict all-cause mortality (10-year risks) were included in the model, by calibrating with BMI (Table). When predicting mortality based on disease mortality only, the risk is fairly constant across BMI categories. However, accounting for combined disease and BMI mortality or BMI mortality alone yields greater differences in mortality risks across BMI categories. 
[_Table Removed - please refer to online abstract for full details_]
**Discussion** 
The updated version of the COM allows the user to apply different ways of calculating mortality. Applying the impact of BMI only on disease-specific case fatality rates results in small differences between BMI categories, whereas applying the effects of BMI on general mortality results in a larger spread of mortality risks. Combining both approaches results in a wider spread of risks, without changing the mortality risks in the normal BMI group, aligned with the evidence base informing the calibration.

**References** 
1. Lopes S et al . Obes Sci Pract 2021;7:269–80. 
2. Lopes S et al . Pharmacoeconomics 2020;38:1123–33. 
3. Bhaskaran K et al. Lancet Diabetes Endocrinol 2018;6:944–53. 
4. Wade K et al. Obesity (Silver Spring) 2018;26:1796–806.

Exploring different ways to predict mortality in a cost-effectiveness model for obesity