Table 2: Cox Regression Models Predicting Mortality at 20-year followup by BMI, and age.

Full sampleAge groups
75–84 year olds85–94 years old
CD and NH residentsCDCD and NH residentsCDCD and NH residentsCD
HRCIHRCIHRCIHRCIHRCIHRCI

Background
Age 1.055–1.081 1.052–1.080 1.044–1.105 1.043–1.105 1.025–1.100 1.025–1.107
Gender (male=1) 1.151–1.492 1.138–1.502 1.266–1.777 1.272–1.8131.093.887–1.3461.049.834–1.318

Origin
 East.857.722–1.017.874.729–1.048.972.777–1.2161.006.799–1.268 .559–.955 .533–.958
 West1.032.889–1.1991.043.888–1.2251.027.849–1.2441.038.848–1.2701.028.805–1.3131.034.786–1.359
Years of education.995.983–1.008.996.983–1.010.997.980–1.015.999.981–1.017.992.974–1.011.993.972–1.015
Additional income.940.824–1.071.933.812–1.072 .711–1.003.858.718–1.0261.040.845–1.279.993.791–1.247
Had children .568–.868 .535–.863 .501–.861 .464–.835.791.554–1.129.780.506–1.204

BMI
Obese (>30)1.117.930–1.3421.091.898–1.326 1.026–1.639 1.036–1.683.944.703–1.268.854.616–1.185
Underweight (<22) 1.202–1.665 1.183–1.677 1.014–1.611 1.036–1.684 1.162–1.870 1.075–1.824
χ 2 ( )

.
.
CD= Community Dwellers; NH= Nursing Home; HR= Hazard Ratio= Exp (b); CI= Confidence Interval.