Table 5. Hazard ratios and 95% confidence intervals for risk of death according to BMI category (40.0–59.9 versus 18.5–24.9 kg/m2) or continuous values of BMI between 40.0 and 59.9 kg/m2 (per 5 kg/m2): subgroup analyses.
| Subgroup | Category | Categorical BMI | Continuous BMI | |||
| BMI 18.5–24.9 kg/m2 (n = 304,011) | BMI 40.0–59.9 kg/m2 (n = 9,564) | p-Interaction* | Per 5 kg/m2 for BMI 40.0–59.9 kg/m2 (n = 9,564) | p-Interaction* | ||
| Overall | Number of deaths | 17,047 | 1,036 | 1,036 | ||
| HR (95% CI)a | 1.00 (reference) | 2.57 (2.41–2.74) | 1.40 (1.31–1.51) | |||
| By sex | Men | |||||
| Number of deaths | 5,404 | 212 | 212 | |||
| HR (95% CI)a | 1.00 (reference) | 2.83 (2.45–3.26) | 1.54 (1.32–1.81) | |||
| Women | ||||||
| Number of deaths | 11,643 | 824 | 824 | |||
| HR (95% CI)a | 1.00 (reference) | 2.53 (2.35–2.73) | 0.03 | 1.37 (1.26–1.49) | 0.01 | |
| By attained age | Follow-up before age 65 y | |||||
| Number of deaths | 3,830 | 368 | 368 | |||
| HR (95% CI)a , b | 1.00 (reference) | 3.10 (2.77–3.47) | 1.59 (1.42–1.79) | |||
| Follow-up after age 65 y | ||||||
| Number of deaths | 13,217 | 668 | 668 | |||
| HR (95% CI)a , c | 1.00 (reference) | 2.34 (2.16–2.53) | <0.001 | 1.30 (1.18–1.43) | <0.001 | |
| By calendar year | Follow-up before 2000 | |||||
| Number of deaths | ||||||
| HR (95% CI)a , d | 1.00 (reference) | 2.50 (2.28–2.76) | 1.35 (1.21–1.51) | |||
| Follow-up after 2000 | ||||||
| Number of deaths | ||||||
| HR (95% CI)a , e | 1.00 (reference) | 2.48 (2.27–2.71) | 0.50 | 1.45 (1.32–1.60) | 0.34 | |
| By geography | US cohorts only | |||||
| Number of deaths | 15,770 | 999 | 999 | |||
| HR (95% CI)a | 1.00 (reference) | 2.60 (2.44–2.78) | 1.39 (1.29–1.50) | |||
| Non-US cohorts only | ||||||
| Number of deaths | 1,277 | 37 | 37 | |||
| HR (95% CI)a | 1.00 (reference) | 1.88 (1.34–2.63) | 0.03 | 1.71 (1.17–2.50) | 0.04 | |
*Calculated by including a cross-product term between the subgroup variable and BMI (40.0–59.9 versus 18.5–24.9 kg/m2); subgroup analyses for attained age and calendar year were stratified by follow-up period.
Models use attained age as the underlying time metric, and are adjusted for sex, race/ethnicity (white [all participants in the Swedish and Australian cohorts were coded as white], black, Asian/Pacific Islander, Hispanic, other/unknown), education (high school or less, post–high school, college, unknown), alcohol intake (grams of ethanol per day: 0, >0 to <10, ≥10, unknown), physical activity level (cohort-specific tertiles corresponding to low, medium, and high), and study.
Models excluded participants who entered the study at or after age 65 y; follow-up time was censored at age at study exit (due to loss to follow-up, death, or administrative end date) or age 65 y, whichever came first.
Models excluded participants who exited the study at or before age 65 y; follow-up began at age 65 y for those who entered the study before age 65 y.
Models excluded participants who entered the study on or after the year 2000; follow-up time was censored at study exit (due to loss to follow-up, death, or administrative end date) or the year 2000, whichever came first.
Models excluded participants who exited the study on or before the year 2000; follow-up began in the year 2000 for those who entered the study before the year 2000.