Table 1.
Effects of successively stricter precautions against bias on analyses of six WHO defined groups of BMI versus all-cause mortality
| Underweight (15·0– <18·5 kg/m2) | Normal weight (18·5– <25·0 kg/m2) | Overweight (25·0– <30·0 kg/m2) | Obesity grade 1 (30·0– <35·0 kg/m2) | Obesity grade 2 (35·0– <40·0 kg/m2) | Obesity grade 3 (40·0– <60·0 kg/m2) | |
|---|---|---|---|---|---|---|
| Crude analysis with no exclusions* | ||||||
| Participants/deaths | 292 003/68 455 | 5 586 892/810 838 | 3 467 617/526 098 | 946 257/144 871 | 237 223/36 113 | 92 458/15 399 |
| HR (95% CI) | 1·82 (1·74–1·91) | 1·00 (0·98–1·02) | 0·95 (0·94–0·97) | 1·17 (1·16–1·18) | 1·49 (1·47–1·51) | 1·95 (1·90–2·01) |
| Participants without known disease at baseline† | ||||||
| Participants/deaths | 255 000/52 789 | 4 922 817/631 488 | 2 916 978/388 781 | 756 075/102 315 | 183 689/24 556 | 696 88/10 321 |
| HR (95% CI) | 1·81 (1·72–1·91) | 1·00 (0·98–1·02) | 0·95 (0·95–0·96) | 1·18 (1·16–1·20) | 1·52 (1·48–1·55) | 2·05 (1·98–2·13) |
| Participants without known chronic disease at baseline, adjusted for smoking status‡ | ||||||
| Participants/deaths | 245 080/51 170 | 4 751 019/618 881 | 2 826 687/381 617 | 733 108/100 113 | 178 130/23 945 | 67 593/10 002 |
| HR (95% CI) | 1·70 (1·61–1·80) | 1·00 (0·98–1·02) | 0·99 (0·98–1·00) | 1·25 (1·23–1·27) | 1·63 (1·59–1·66) | 2·24 (2·15–2·33) |
| Participants without known chronic disease at baseline, adjusted for smoking status, and excluding the first 5 years of follow-up§ | ||||||
| Participants/deaths | 208 044/33 817 | 4 234 052/496 310 | 2 513 128/312 450 | 641 237/80 037 | 152 741/18 737 | 56 232/7 659 |
| HR (95% CI) | 1·60 (1·51–1·70) | 1·00 (0·98–1·02) | 1·03 (1·01–1·04) | 1·31 (1·29–1·33) | 1·70 (1·67–1·74) | 2·36 (2·27–2·45) |
| The primary prespecified analysis: never-smokers without known chronic disease at baseline—excluding the first 5 years of follow-up¶ | ||||||
| Participants/deaths | 114 091/12 726 | 2 145 550/192 523; | 1 250 103/130 293; | 330 840/37 318 | 80 827/9 179 | 30 044/3 840 |
| HR (95% CI) | 1·47 (1·39–1·55) | 1·00 (0·98–1·02) | 1·11 (1·10, 1·11) | 1·44 (1·41–1·47) | 1·92 (1·86–1·98) | 2·71 (2·55–2·86) |
CIs were calculated with floating variance estimates (reflecting independent variability within each group, including the reference group). Reference group is normal weight (18·5–<25·0 kg/m2). All analyses are adjusted for age and sex. Baseline BMI categories were defined by WHO. BMI=body-mass index. HR=hazard ratio.
*
237 studies; 10 622 450 participants; 1 601 774 deaths.
†
236 studies; 9 104 247 participants; 1 210 250 deaths.
‡
234 studies; 8 801 617 participants; 1 185 728 deaths.
§
213 studies; 7 805 434 participants; 949 010 deaths.
¶
189 studies; 3 951 455 participants; 385 879 deaths.