Stewart, Cutler, and Rosen 2009 — Forecasting the Effects of Obesity and Smoking on U.S. Life Expectancy

Summary

Stewart, Cutler, and Rosen use life-table simulation to forecast 2005–2020 U.S. life expectancy (LE) and quality-adjusted life expectancy (QALE) for a representative 18-year-old, given simultaneous trends in rising body mass index (BMI, +0.5%/year) and declining smoking (−1.4%/year). The BMI increase (−1.02 LE years) overwhelms the smoking benefit (+0.31 LE years), producing a net combined effect of −0.71 LE years and −0.91 QALE years over the forecast window. The paper validates the model against 1990–2004 historical trends (actual gain: 2.44 years vs. 2.98-year no-change counterfactual) and shows that full elimination of both risk factors would yield +3.76 LE years and +5.16 QALE years.

Key Claims

• Smoking declined at ~1.4%/year over the 15 years before 2005; BMI increased at ~0.5%/year over the same window.
• By 2020, ~45% of the U.S. adult population is projected to be obese (vs. ~30% in 2005), extrapolating 15-year trends.
• Smoking decline alone: +0.31 LE years, +0.41 QALE years by 2020 (holding BMI fixed at 2005 levels).
• BMI increase alone: −1.02 LE years, −1.32 QALE years by 2020 (holding smoking fixed at 2005 levels).
• Combined (base-case, both trends continuing): −0.71 LE years, −0.91 QALE years by 2020.
• The negative effect of obesity persists as long as annual BMI growth exceeds 0.15%/year — a very low threshold given the 0.5%/year observed rate.
• Full counterfactual (eliminate both risk factors by 2020): +3.76 LE years, +5.16 QALE years; smoking elimination alone: +1.73/+2.17; BMI normalization alone: +1.40/+2.44.
• Historical validation: 1990–2004 LE rose 2.44 years; the no-change-in-risk-factors counterfactual would have produced +2.98 years — adverse trends already reduced gains by ~0.54 years during this pre-forecast period.
• Quality-of-life gap between morbidly obese smokers and normal-weight never-smokers (age 65+) is ~0.17 EQ-5D points (0.61 vs. 0.78).

Concepts Introduced or Extended

• Obesity and Life Expectancy
• Disability-Free Life Expectancy
• SSA Mortality Forecasting

Entities Mentioned

• Susan Stewart
• David Cutler
• Allison B. Rosen
• S. Jay Olshansky

Quotes

"Unless effective strategies to reduce obesity are developed, the negative effect of obesity on life expectancy will increasingly offset the positive effect of declining smoking rates."

"We project that if recent trends in the prevalence of obesity and smoking continue, they will have a combined effect of reducing life expectancy by 0.71 years by 2020."

My Take

The paper's central contribution is methodological transparency: by decomposing smoking and BMI effects separately before combining them, it demonstrates that the obesity drag is approximately 3× larger than the smoking benefit — making the net direction of LE change in the medium term a function of which trend dominates. The threshold finding (0.15%/year BMI growth) is particularly policy-relevant: it shows there is essentially no "slow-growth" scenario under which obesity has a neutral effect unless BMI growth actually stops. The historical validation (1990–2004) is credible, though the counterfactual implicitly assumes no other risk-factor changes. The paper does not model race/socioeconomic-status (SES) heterogeneity, meaning the forecasted LE drag is likely worse for lower-SES groups where obesity rates are rising fastest. Also directly relevant to SSA Mortality Forecasting: this paper's framework demonstrates why LE projections ignoring obesity and smoking covariates (as the Social Security Administration's (SSA) Office of the Chief Actuary (OACT) historically did) are systematically overoptimistic.